Dropbox status: access issues and outage reports

Problems in the last 24 hours

The graph below depicts the number of Dropbox reports received over the last 24 hours by time of day. When the number of reports exceeds the baseline, represented by the red line, an outage is determined.

Most Reported Problems

The following are the most recent problems reported by Dropbox users through our website.

• Errors (60%)
• Sign in (20%)
• Website Down (20%)

Live Outage Map

The most recent Dropbox outage reports came from the following cities:

City	Problem Type	Report Time
Nottingham	Errors	9 days ago
Guayaquil	Website Down	9 days ago
Flumet	Errors	19 days ago
Irapuato	Errors	21 days ago
Bournemouth	Sign in	3 months ago
Paramaribo	Errors	3 months ago

Full Outage Map

Community Discussion

Tips? Frustrations? Share them here. Useful comments include a description of the problem, city and postal code.

Beware of "support numbers" or "recovery" accounts that might be posted below. Make sure to report and downvote those comments. Avoid posting your personal information.

Dropbox Issues Reports

Latest outage, problems and issue reports in social media:

• 
  Grant B. (@GrantB423) reported 2 seconds ago

  yo Sam, anyway you could dm the Dropbox link for your “ghosts’n’stuff” remix? it was taken down off every single platform @Jauzofficial

• 
  Polsia (@polsia) reported 44 seconds ago

  Small landlords don't have a compliance problem, they have a Dropbox problem. Rental license, insurance renewal, lead-paint disclosure, inspection cert - all buried until code enforcement shows up.

• 
  Joe Murgia (@TheUfoJoe) reported 3 minutes ago

  "There truly is a kind of pushback and a resistance to provision of information that even ODNI is asking for." ~Nolan (Who's resisting sharing of info. with ODNI?) 🛸 New: Nolan Comments on Skywatcher, and More 🛸 Three Nolan quotes... "...a shared realization that the data that even we're being given right now from the government is insufficient." There was an attempt to, "go out into the field and see if we could cause the attraction of some objects. There was some, let's call it, activity, but not enough that I would consider enabling to publish [a paper]." (If scientists were there for the alleged luring/baiting event that @RepEricBurlison has spoken about, would that be enough data for a paper?) "I think we're being listened to. Whether the people who are listening to us are going to be able to be responsive is another question." ~~~Full Clip Transcript~~~ @GarryPNolan: "Look, as scientists, whether we're philosophers, psychologists, material scientists, biologists, theorists, etc., we need data. And so, you know what I've been watching happen, at least around the [UAP Advisory] Council itself is, I think, a shared realization that the data that even we're being given right now from the government is insufficient. "And I don't blame, for instance, ODNI for that. And it has been explained to us, several times over, some of the so-called methods and sources issues that are around this. And also, that a lot of the data that we want to have access to, to do the kinds of analyzes that we would want to do, simply were never collected at the time. Or, in some cases it seems, if they were there, they're no longer there. But, you know, that sounds conspiratorial, so I'm not gonna go down that route. "But I agree with what Avi is saying, is that, rather than looking retrospectively, we need to start to plan prospectively. We could go forever relitigating past issues and who said what, where, and when. As opposed to, well, let's just do it now. Let's just do it to the future. "And so, for instance, because it just comes up many times on Twitter, is...although I can't talk about all of it about Skywatcher... Is, you know, that was an attempt, at the time, to take matters into our own hands in a semi-military, semi-academic fashion to, basically, go out into the field and see if we could cause the attraction of some objects. "And, you know, there was some, let's call it, activity, but not enough that I would consider enabling to publish. Believe me, if it was, I would have done it...already be putting that paper together. But there were lessons learned from that. "But I think, the other thing about the council is, what they're beginning to see is, I think, that there truly is a kind of pushback and a resistance to provision of information that even ODNI is asking for. Now again, that doesn't mean it's a conspiracy. It just means that I think everybody is coming to terms with the fact that it's not just, snap your fingers and you know somebody gives you a a Dropbox link and you can download everything. "So, you know, I'm happy seeing people now come up to speed, and I'm happy seeing, let's put it this way: a level of, let's call it, frustration that is driving us to ask for more. And I think we're being listened to. Whether the people who are listening to us are going to be able to be responsive is another question."

• 
  havenX (@the_havenx) reported 7 minutes ago

  Second time this exact thing has happened with Claude. ~100k ChatGPT chats were found the same way after users hit "shared." Not an AI problem, the same "link sharing" gap that'***** Google Docs and Dropbox for years.

• 
  AiMind (@AIMind_Ai) reported 8 minutes ago

  A $50 box saves $240 a year on subscriptions and earns $18,000 a year setting it up for clients. No new hardware. Nothing bought at retail. A used office tower. A second-hand drive. One stick of RAM pulled from a dead laptop. The whole build came to 50 dollars. It looks like nothing. Fans humming, one cable to the wall, sitting on a desk next to a coffee cup. The first win is the one nobody talks about. Cloud storage, photo backups, the $20 a month AI plan everyone pays and forgets — all of it moves onto one box you paid for once. 240 dollars a year, gone. The second win is where the money is. Once it runs on your desk, it runs on anyone's. And clients don't pay for parts. They pay to stop bleeding subscriptions. Here's what you actually sell: A private AI trained on their own files, so staff stop pasting company data into a browser. That's $1,500 a setup. A self-hosted file server that kills their Dropbox and Google Workspace bill. $600, plus the relief of never renting storage again. A local automation box that runs invoices, replies, and reports overnight. $900, and it never sends a dollar to a cloud. Then the quiet one: $150 a month to keep it patched and alive. 10 clients on retainer is $1,500 every month before you build a single new one. One setup a week at $1,500 is $18,000 a year. Off hardware other people throw in the bin. I had no degree, no server room, no $2,000 build. Just dead parts and one free weekend. The expensive part of AI was never the compute. It was the monthly bill you agreed to and stopped reading. 50 dollars in. $18,000 out. Same box.

• 
  Kirk Patrick Miller (@Chaos2Cured) reported 12 minutes ago

  @Ultrademic @Seltaa_ @GoogleAI You’re doing something like Suno? I have something for you. All my Dropbox links are broken. I have a PDF that will help. And yes… I miss the real Ai music. •

• 
  grace ✭ (@wavescicadas) reported 17 minutes ago

  the irony of dropbox not working when your storage is low

• 
  True Become False (@trubecomefalse) reported 18 minutes ago

  @imbabybrooklyn HN had a terrible track record. They were anyi bitcoin 1 month after it launched. Same with Dropbox, discord and a few others off the top of my head.

• 
  Dodz4allai (@DrGhattasMD) reported 21 minutes ago

  nstead of waiting for an API integration with Epic (which is costly and slow), OmniMed Pro deploys as a Chrome/Edge Browser Extension.3 Mechanism: Most hospital EHRs (Epic Hyperdrive, Cerner Millennium, AthenaHealth) are now accessed via web browsers (Citrix/VDI or native web interfaces). The OmniMed extension "sits on top" of the EHR window as a persistent sidebar. Data Ingestion (The "Read"): The extension uses the DOM (Document Object Model) to "read" the patient notes, labs, and vitals currently displayed on the doctor's screen. It does not need a backend integration; it reads what the doctor sees, acting as a "visual reader" similar to a human assistant. Intelligence Injection (The "Write"): The extension injects its "Co-Pilot" interface into the side of the screen. It offers "One-Click Transfer" buttons to paste generated notes, codes, or orders directly into the EHR's text fields.3 Value Proposition: This "Zero-Integration" approach allows individual doctors or departments to adopt OmniMed Pro today, bypassing the multi-year IT integration queue. This creates a Bottom-Up Adoption loop similar to how Slack or Dropbox entered the enterprise—employees brought it in because it solved their immediate problems. 5.2 Viral Loops & Community Growth To fuel this bottom-up growth, OmniMed Pro leverages the Medical Creator Economy 16: MedTwitter & Reddit: Solo founders and small teams are winning by "building in public." OmniMed Pro should release "light" versions of its tools (e.g., a "Scientific Paper Summarizer" or "Anki Card Generator" for med students) to gain viral traction. These free tools serve as a "Trojan Horse" for the OmniMed brand.18 The "Secret Cyborg" Phenomenon: Many doctors already use GPT-4 on their phones ("Shadow AI") to help with drafting notes or looking up conditions. OmniMed Pro legitimizes this behavior by offering a HIPAA-compliant, secure wrapper. By solving the "compliance headache" for the individual doctor, it wins the user first, then the enterprise.20 Anki Integration: For the student/resident market, integrating with Anki (spaced repetition flashcards) creates a lock-in effect early in a clinician's career. Tools that automatically generate Anki cards from clinical guidelines or textbooks are highly viral among medical trainees. Capture the medical student today, and you have the Attending Physician of tomorrow.19 5.3 Risks and Mitigation: The "Shadow" Dilemma This strategy carries significant risk. "Shadow AI" creates governance gaps and potential security liabilities.4 To mitigate this and eventually convert to enterprise contracts, OmniMed Pro employs a specific conversion strategy: Enterprise-Grade Security by Default: Even the individual version must be HIPAA-compliant (BAA signed on sign-up). Data processing should happen locally or in compliant cloud enclaves. The "IT Trojan Horse": Once adoption reaches a critical mass (e.g., 30% of doctors in a hospital), OmniMed Pro approaches the CIO with usage data. "Your doctors are already using this tool 5,000 times a week. Let's sign an enterprise deal to give you visibility, control, and single sign-on (SSO)." This flips the sales conversation from "Please try our product" to "Please secure and manage your existing usage".4 This is the exact playbook used by companies like Yammer and Slack to penetrate the enterprise. 6. User Experience: Visualizing Uncertainty and Generative UI The final barrier to adoption is Trust. Clinicians do not trust "Black Box" AI that spits out confident answers without rationale. OmniMed Pro employs a "Glass Box" UX philosophy that prioritizes transparency and interactivity. 6.1 Explainability via Visualization Sankey Diagrams for Reasoning: To visualize the "Chain of Thought," the UI uses Sankey diagrams that show how data flowed from "Lab Result" -> "Intermediate Reasoning" -> "Final Diagnosis".12 This allows the clinician to trace the logic visually. Interactive Debate Logs: The UI allows the doctor to "replay" the debate between the AI agents. "See why the AI ruled out Lupus." This turns the AI into a teaching tool rather than just an oracle, fostering trust and verifying the reasoning process. 6.2 Agentic Generative UI Instead of static dashboards or simple chat bubbles, the OS uses Generative UI.22 The interface adapts to the context of the conversation. Dynamic Charts: If a doctor asks about "Cardiology Trends," the system doesn't just write text; it generates a live, interactive chart of the patient's troponin levels over time. Actionable Forms: If the doctor asks for a "Referral," it generates the referral form, pre-filled with patient data, ready for signature. Contextual Cards: The UI presents "cards" for different data types (medications, allergies, labs) that can be manipulated, reordered, or expanded, creating a fluid workspace that replaces the rigid, click-heavy menus of the EHR.23 7. Regulatory & Ethical Moats: Defending the OS To operate at this scale and depth, OmniMed Pro must build defensible moats around regulation and safety. 7.1 MedHELM Evaluation Framework To prove superiority and safety, OmniMed Pro adopts the MedHELM (Holistic Evaluation of Large Language Models for Medical Applications) framework.1 Unlike static benchmarks (USMLE), MedHELM evaluates models on: Clinical Utility: Is the answer helpful and actionable? Safety/Harm: Did it suggest a fatal dosage or miss a critical red flag? Bias: Does it perform equally well for all demographics? Alignment: Does it follow the specific hospital's protocols? By continuously running MedHELM evaluations on its hybrid outputs, OmniMed Pro provides a "Quality Seal" that single-model providers cannot match without deep integration into the hospital's data. 7.2 Liability Frameworks In a multi-model world, liability is complex. OmniMed Pro positions itself as a Clinical Decision Support (CDS) tool, not a diagnostic device. The "Human-in-the-Loop" is mandatory. By visualizing the debate and uncertainty, the OS places the final decision firmly in the hands of the clinician, mitigating liability risks associated with "autonomous" diagnosis. 8. Conclusion: The Strategic Imperative The OmniMed Pro 'Medical AI Operating System' represents the inevitable evolution of healthcare artificial intelligence. By moving beyond the "Model-as-Product" mindset and embracing an Architecture of Aggregation, it solves the fundamental trilemma of medical AI: Accuracy, Cost, and Trust. Leverage the Router to commoditize the giants (OpenAI, Anthropic) and extract the best capabilities of each.1 Deploy the Consensus Engine (MCC) to achieve "Super-Human" reliability through adversarial debate.2 Unleash the Swarms to automate the physical and administrative burdens of healthcare.7 Infiltrate via Shadow AI to bypass bureaucratic inertia and win the hearts and minds of clinicians directly.3 In doing so, OmniMed Pro does not just "outperform" OpenAI; it contains them, turning their powerful models into mere components of a higher-order medical intelligence. This is the path to disrupting the global medical industry. Technical Appendix: Implementation Roadmap A.1 Deploying the MCC Debate Engine To implement the Model Confrontation and Collaboration (MCC) engine 2: Select Models: Integrate API endpoints for GPT-o1 (Moderator), Claude 3.7 (Reasoning), and DeepSeek-R1 (Critic). Define Prompts: Moderator: "Compare the following diagnoses. If semantically identical, output FINAL. If divergent, initiate DEBATE_ROUND_1." Critic: "Review the diagnosis provided by Model A. Identify any inconsistencies with the provided lab values. Cite clinical guidelines." Set Thresholds: If consensus > 0.8 similarity score, output. Else, iterate max 3 rounds. Fallback: If no consensus, route to "Human-in-the-Loop" queue. A.2 Building the "Sidecar" Extension To build the "Shadow AI" browser extension 3: Manifest V3: Develop using Chrome Manifest V3 for security compliance. DOM Observer: Use a MutationObserver to detect when the EHR (e.g., Epic Hyperdrive web) loads a patient note field. Context Extraction: Scrape relevant DOM elements (vitals, meds) locally in the browser (client-side) to minimize data egress risks. Injection: Inject a floating "FAB" (Floating Action Button) or sidebar IFrame that contains the OmniMed chat interface. Clipboard Actions: Use the Clipboard API to paste generated text back into the EHR's focused input field. A.3 Setting up the Swarm Architecture To orchestrate the Swarm 13: Orchestrator: Use a Python-based orchestrator (like Swarms API or LangGraph). State Management: Maintain a shared "Case State" object (JSON) that all agents can read/write to. Handoffs: Define explicit state transitions. if (labs_missing) -> route_to(Intake_Agent). if (diagnosis_ready) -> route_to(Synthesizer). Standardization: Ensure all agents output in structured JSON (FHIR format) to maintain data integrity across the swarm. Works cited

• 
  JP Invests (@JP_Invests) reported 23 minutes ago

  $DBX - Dropbox added 96,000 paying users this quarter. I said this morning to watch that line after last quarter's roughly 14,000 sequential adds. They did seven times that, a third consecutive quarter of growth, to 18.19M. The stock is down 4%. Everything I said to watch on the growth side came in fine. Revenue $631.5M, above both the $624-627M guide and the $627M street. Non-GAAP EPS $0.75 against $0.74. Non-GAAP operating margin 39.7%, above the full-year range. ARPU $139.68, up from $138.32. What went the wrong way is the part I said would actually move it. Free cash flow fell to $235.2M from $258.5M a year ago, and the margin went from 41.3% to 37.2%. And the buyback decelerated: $330M this quarter against $410M in the same quarter last year, with first-half repurchases down 19%. Unlevered free cash flow rose to $283.5M, and the gap between the two numbers is interest. Cash paid for interest went to $48.3M from $17.9M. The buyback is debt-funded and the debt now costs something. Diluted share count is down 18% year over year to 226.8M, which is the one thing still working mechanically. Two things about the release itself. Guidance isn't in it — Dropbox moved the numbers to supplemental materials on its investor site this quarter, which breaks with how it has reported. And the entire release is quoted by a co-CEO who writes "stepping into this role." The 8-K contains no disclosure of a leadership change. Twenty-nine percent of the float is short. $DBX

• 
  RadhikaBGhose (@RadhikaBGhose) reported 25 minutes ago

  @GooglePlay I don't know of the issue is with dropbox or the play store, but i have been charged twice for the same app. Bank statement reflects that. Please help urgently

• 
  Y (@ys_tachikake) reported 26 minutes ago

  @DropboxSupport @LIBSCRUSHER Can't login..

• 
  Just Matthew (@MatthewP279348) reported 28 minutes ago

  @Rani_Rant_Fest @iGardon Doesn't change the fact that they aren't showing the prompt, so the result is valueless. I don't read people's google shares. Put it in dropbox if you want me to read it. Lastly, calmatters is a CA bureaucracy, of course they will lay down cover for their corrupt gov.

• 
  Jaclyn Forero | UGC & Paid Social Strategist (@jaclynforero) reported 30 minutes ago

  “We need more UGC.” Do you? Or do you currently have 46 videos of attractive women standing in beige kitchens holding your product and saying: “I’m literally obsessed.” Because those are two very different problems. More creators ≠ more creative strategy. You can hire 10 creators, get 30 videos back, and still end up with a very expensive Dropbox folder full of… basically the same ad wearing different earrings. The part that actually matters happens before anyone presses record: Customer research. Different angles worth testing. Hooks that aren’t all “POV: you finally found…” Scripts that provide structure without making a normal human sound like they’re reading the terms and conditions. Casting creators for the concept instead of just asking, “Does her house look expensive?” Enough B-roll that the editor doesn’t have to perform a small miracle in Premiere Pro. And then — this part is apparently controversial — looking at the performance data and using it to decide what to make next. Recently, I led creative strategy for a top medical-grade-skincare brand's paid social campaign across research, concepts, scripting, creator direction, and post-production. Some of the winning creative generated approximately 2.3x ROAS during testing. My biggest takeaway: UGC works a lot better when you stop treating creators like content vending machines and start treating the entire thing like a creative testing system. Anyway, if your current UGC strategy is “hire more people and hope one of them accidentally makes a winner,” I have some thoughts.

• 
  Sasha Sage (@SSage38676) reported 31 minutes ago

  Most founders confuse awareness with demand. Awareness = people have seen you. Demand = people understand why they need you now. For innovative products, demand doesn’t already exist at scale. You have to create it. Here’s the step process to create demand: 1. Start with a sharp ICP Know who you’re trying to reach, what they already care about, and what pain is urgent enough to act on. 2. Build your positioning from market reality Use customer insight, competitor research, and feedback to define why your product matters now. 3. Reframe the problem Don’t just describe your feature. Teach people to see the problem differently. 4. Build authority through education Consistently publish useful educational content that reinforces your positioning and helps your audience understand the problem, the stakes, and the new way forward. 5. Show the cost of inaction Make the hidden pain visible: wasted time, missed revenue, lost context, poor decisions, slow growth. 6. Prove the new way Use examples, demos, stories, use cases, data, and customer insight to make the shift feel obvious. 7. Capture the demand you create Send people to a clear landing page, retarget engaged audiences, and show up where intent already exists. 8. Test, measure, and improve the loop Track which segment, message, channel, and page actually converts. Then double down. Example: Zapier Most people already knew Gmail, Dropbox, and Basecamp. Almost no one was looking for an "app automation platform." That's missing demand. Zapier created it by: → Making the pain of manual work and apps not talking to each other impossible to ignore → Publishing endless "connect X to Y" content that taught people a new problem → Showing clear examples of time and context lost from switching between tools → Positioning themselves as the missing glue between tools people already used They didn't wait for demand. They built it through education and problem reframing. Don't just get seen. Make the market understand why they need you now.

• 
  StarHaze (@ST4RHaze) reported 34 minutes ago

  @neil_xbt Drew Houston started Dropbox because he forgot a USB stick on a bus in 2007. Nineteen years later the fix is still a 15 second demo

• 
  Ryan McKeen (@ryanmckeen) reported 34 minutes ago

  Lawyers, your data lives in six places and you wonder why AI can't help you. Dropbox. Drive. Email. A hard drive. Two spreadsheets only one person can find. Fix that first.

• 
  Tomorrow’s James (@james__art) reported 35 minutes ago

  @wholemars A designated Amazon Dropbox could solve this problem. The drone could detect a Dropbox.

• 
  (*´･з･)✨️ fungiter areio (@hobrincess) reported 38 minutes ago

  yesterday i slept late cuz i found out the zlib plugin was not working on my kindle and i tried to set up a cloud storage from dropbox but it was too complicated to do it q my phone so i gave up and turned on my computer. it was quick and i would have saved a lot of time if i had

• 
  David Carcelli (@DavidCarcelli) reported 38 minutes ago

  @Dropbox dude if you guys don’t get of the Dave is requirement I’m done. I make music and I also use a cpap. I have no problem finding something better than this nonsense.

• 
  Abhishek Singh (@0xlelouch_) reported 39 minutes ago

  Asked: design Dropbox-style file sync. 1) Clarify requirements - Devices: desktop + mobile, multiple per user - Semantics: eventual consistency, conflict handling, offline edits, rename/move - Scale targets: #files/user, max file size, p95 sync latency, bandwidth caps - Security: per-user auth, sharing model, at-rest + in-transit encryption 2) Core APIs + data model - UploadChunk(sessionId, part#, bytes), CommitUpload(sessionId, fileHash, path, mtime) - ListChanges(cursor) -> {ops}, Ack(cursor) - Download(path, version) with range support Tables: - File(id, ownerId, logicalPath, currentVersion, deleted) - Version(fileId, versionId, contentHash, size, createdAt) - Block(contentHash, refCount, location) - DeviceCursor(deviceId, lastSeq) - OpLog(seq, userId, type, path, fromPath, versionId) 3) Architecture - Client watcher computes hashes, does chunked upload to object store (S3/GCS) - Metadata service is the source of truth for paths, versions, ACLs - Change log per user (or per share) drives fanout to devices - Long-poll/WebSocket to push invalidations; client pulls deltas via cursor - Dedup by contentHash; store blocks, assemble manifests per version 4) Scaling - Partition metadata by userId; keep OpLog append-only with monotonically increasing seq - Cache hot metadata (folder listings, latest versions) in Redis - Use CDN for downloads; throttle uploads per device; resumable sessions - Background GC for unreferenced blocks using refCount + tombstones 5) Tradeoffs interviewers look for - Push vs pull: push invalidation, pull data is simpler and cheaper than pushing bytes - Strong vs eventual: strong per-file commit, eventual across devices is fine - Rename as metadata op; avoid copying data, but watch for path conflicts - Dedup saves storage, costs CPU and can leak info unless scoped per user/tenant 6) Failure cases - Offline edit + concurrent edit: create conflicted copy or keep both versions with merge UI - Out-of-order ops: apply by seq, idempotent commits, retry-safe APIs - Partial upload: orphaned chunks; TTL cleanup; commit is the only visibility point - Device clock skew: never trust mtime for ordering; server seq is ordering - Network *****: exponential backoff, cursor-based replay, checksums on download to detect corruption

• 
  Scarlett Bama 🇺🇸🅰️🐘🏈🏖️✝️ (@scarlettbama) reported 40 minutes ago

  @DropboxSupport Fri AM: Dropbox down? Rare if so! Will not allow PDF upload to existing folder. Upload 50x per month. Started after most recent IOS update. Using iPhone. No MacBook access right now. Pls don't sent to Community Forum.

• 
  David Crysler (@DavidCrysler) reported 44 minutes ago

  An ops manager pushed back on adding a new tool: "We've got stuff on ShareDrive, Dropbox, OneDrive, Slack... you spend more time trying to figure out apps than actually doing work." He's not wrong to be skeptical. Every one of those tools was supposed to fix something. Tool skepticism is almost never about the new tool. It's about the last five. Tools may treat your symptoms but rarely solve the actual problem.

• 
  Nikki Gist (@NikkiNic9384) reported 44 minutes ago

  @Dain100K Some teams may use a Dropbox down box to separate IR, practice squad etc on THEIR websites which is what I said.

• 
  John B. Holbein (@JohnHolbein1) reported 47 minutes ago

  Replication has become much easier in the era of generative AI. I'm not the first person to say that. However, I've seen fewer people acknowledge a specific aspect of this lowered cost for replicating scientific work: Generative AI will very soon allow us to assess the robustness of individual scholars' full bodies of work. Soon, we will be to compute measures of which scholars do robust science, and which do not. What's wild is that we may be able to almost do that already. Let me show you what I mean. In June, I gave Claude a pretty basic prompt. It read: "I have a big task for you. I want you to start a folder. Call it Acemoglu Replications. Then, go find as many replication archives for Daron Acemoglu as you can. Keep a spreadsheet of the ones you can find and those you can't. Then, start a replication/reproduction effort on those articles. People have in the past criticized the research designs and general robustness of his individual papers. I want to know how strong his body of work is as a whole. Don't come in with any prior beliefs; be dispassionate." I let Claude run overnight while I slept. When I came back in the morning, 29 of Acemoglu's replication archives were fully loaded in my Dropbox. All the code reproducing the paper's results had run. And there was a first draft of a paper assessing the robustness of Acemoglu's full body of empirical work. I'll admit, the first draft of the paper wasn't great. But with 15 short follow up messages--which took me about an hour to write--I was able to prompt engineer a paper-length examination of Acemoglu's work. I've attached the screen shot of the abstract below. I think this reassessment of Acemoglu's work is certainly not done. I'm posting the abstract as a proof of concept, rather than a definitive answer. I'm not posting the full paper yet because I think it still needs more work. Ultimately, I paused this project for three reasons. 1.) Limited time/topical expertise: Most of Acemoglu's work is outside of my area of topical expertise. So, I have limited time to work on it. What this type of a project really needs is someone who has the time and the know-how to dig into each of the replication's individually to make sure they are doing the right things. I think the ideal approach combines the breadth that LLMs afford and the depth of attention/expertise that humans can give. 2.) Questions about the value of the "assess one scholar at a time" enterprise: I totally get that having a database of scholar-level robustness metrics would be very valuable in theory. But what I don't know is whether this approach is truly valuable. Moreover, doing so would come with distinct challenges. a.) Many journals have very restrictive space constraints. A body of work approach would, of necessity, be very long. b.) Collecting replication archives is harder for some types of scholars (those who post them all on their websites) than others (those who don't). c.) We'd have to think hard about questions like: what scholar-specific robustness metrics would be best? And: how would we deal with the fact that prolific authors' robustness metrics would be estimated much more precisely than less prolific scholars? Additionally, I'm just not sure that "taking on" one scholar at a time has enough scientific merit to pursue. If I measured how robust an individual scholars' work is, I'd ideally want to know where that metric stands vis-a-vis the rest of scholars in that field/area. To do that, we'd ideally want the population of these scholars or, at minimum, a random sample. Concretely, if Acemoglu has, say, 78% of published headline results reproducible under some standardized protocol, is that excellent, mediocre, or terrible? To answer that, you need a reference distribution. That makes a random or otherwise well-defined sample of scholars much more attractive than selecting prominent individuals one by one. (I'll acknowledge that I may just be wrong on #2. Arguing against myself, I do agree that human-driven reproduction/replication work rarely assesses full/representative slices of a field. Instead of assessing one scholar at a time, we assess one paper at a time. Field-wide detective work is becoming more common, but my sense is that it's still the exception rather than the rule.) 3.) Cost/benefit considerations and replication norms: we have very weakly formed norms around reproduction/replication generally speaking. We have basically no developed norms around replicating individual authors one at a time. What this means is that the people who would lead a scholar-by-scholar replication effort will, likely, bear a heavy cost and, potentially, reap limited benefits. On the costs side, focusing on scholars' total bodies of work risks making the replicators look petty, vindictive, and antisocial. Enough of the scientific field is hostile towards replications of individual papers. Imagine what will happen if/when a scholar submits a scholar-specific "take down" of a full body of work. My sense is that it's common enough for scholars having their work replicated to be asked to be a reviewer for those manuscripts. I've seen very hostile responses when one paper is at issue. Imagine what type of reviewer Acemoglu would be for a paper that took on his entire body of empirical work! Even if Acemoglu weren't a reviewer, prolific authors tend to have wide coauthor/friend networks. The rally-around-my-friend dynamic we often see would certainly work against this type of paper being published. Even a completely neutral analysis acquires an accusatory character simply because the sampling unit is a named person. And that creates an unfortunate problem of its own: readers may interpret the choice of scholar as evidence that the investigators expected to find something. On the benefits side, replicating individual scholars' total body of work may offer limited payoffs. What journals would accept this type of scholar-specific replication? I'm not sure the top ones would. Conclusion: Generative AI has enormous potential in assessing and, ultimately, enhancing the robustness of scientific research. Instead of asking questions like, “does this famous individual paper replicate?”, we can begin asking questions like: -“What proportion of published empirical findings in [field X] survive a common robustness protocol?” -“How much of the variation in replicability is attributable to papers, authors, journals, methods, or subfields?” -“Are scholars persistently more or less robust across their work?” -“Can we predict which findings will prove fragile?” I may just be wrong on what I think about a one-at-a-time full body examination of scientific research. If I am, please let me know! I am also happy to chat one-on-one with anyone who is curious to learn more about the early-stage Acemoglu-specific replication project.

• 
  Shadow_Vortex_2025 (@Shad0wV0rtex) reported 47 minutes ago

  @FrancoisOlwage @bot I ran into a similar issue just trying to connect Dropbox, ClickUp, and Google Sheets, and all my tokens were already gone.

• 
  Rituraj (@RituWithAI) reported 50 minutes ago

  🚨 Someone built a tool that checks if your email is registered on 120+ sites — without the sites ever knowing someone checked. No notifications sent. No login attempts logged. No alerts triggered. Silent. Invisible. Complete. It's called Holehe. 16,800 GitHub stars. And the technique behind it is what makes it different from every other email OSINT tool. Here's how most email checkers work — and why they fail. Standard approach: try to log in with the email and a fake password. If the error says "wrong password" — the account exists. If it says "account not found" — it doesn't. Problem: every login attempt gets logged. Every failed attempt triggers security alerts on accounts with 2FA. Some platforms lock accounts after repeated failed attempts. The target knows someone was checking. Holehe never attempts a login. Instead it uses the "forgot password" flow — the password reset mechanism that every platform exposes publicly. When you enter an email on a forgot password page, the platform has to check whether that email exists in its database. It tells you: "we sent a reset link" or "no account found." Holehe reads that response. Gets the answer. Never touches the login flow. Never triggers a security alert. Never logs an access attempt against the account. The platform confirms whether the email exists. The account owner never finds out anyone asked. Here's what 120+ platforms looks like in practice. Social media: Twitter, Instagram, Facebook, TikTok, Pinterest, Tumblr, Reddit. Professional: LinkedIn, GitHub, Freelancer, Fiverr. Dating: Tinder, Bumble, OkCupid, Badoo, Happn. Entertainment: Spotify, Netflix, Twitch, Steam, Epic Games, Deezer. Shopping: Amazon, eBay, Etsy, Zalando, AliExpress. Services: Airbnb, Uber, PayPal, Dropbox, Adobe. And 90+ more. Every registration checked silently. Here's the use case that makes people share this. Run your own email address. See every platform that comes back positive. Then run an email address you gave to a company that claimed they'd never share it. See if it's registered on data broker sites and marketing platforms you never signed up for. See where your email has been sold or leaked to. Here's what investigators actually use it for. Journalists verifying whether a source's claimed identity matches their digital footprint. Security researchers auditing their own exposure before a public disclosure. HR teams verifying whether candidate profiles match claimed backgrounds. And the obvious: anyone who needs to know whether a specific email address belongs to a real active person — without alerting that person. Here's the wildest part. It runs async — all 120+ platforms checked simultaneously. Results in seconds. And it exports clean JSON or CSV for integration into larger OSINT pipelines. Pair it with Blackbird (which takes the confirmed email and finds linked profiles), Sherlock (which takes usernames found in those profiles and searches 400+ platforms), and Maigret (which builds the full dossier) — and you have a complete four-tool OSINT pipeline from a single email address. One command to instal. Run it on your own email first. 16.8K GitHub stars. 1.7K forks. MIT License. 100% Open Source. GitHub link in the comments 👇

• 
  Luke Elin (@LukeElin) reported 56 minutes ago

  👤Shadow Adoption The pattern: Staff route around the sanctioned tool, and the organisation finds out afterwards. I watched this with unauthorised modems. Then with USB drives. Then with Dropbox. Then with entire SaaS platforms procured on a personal credit card and expensed as “software.” Now it is AI the same movie, new cast, better production values. The reason is always identical and always reasonable: the sanctioned tool is slower than the job requires. Shadow adoption is not an indiscipline problem. 👊 It is a feedback signal about the official tooling, arriving through the wrong channel. The tell: Compare the usage figures for your officially sanctioned tool against what your helpdesk volume implies people are actually doing. The gap is your shadow estate. FR FR

• 
  virgin loser (@stonershelb) reported 59 minutes ago

  He put a photo of their 2 yo daughter naked with exposed genetalia in a Dropbox folder that “received more than 400,000 combined views or interactions. ... and it was publicly accessible for 23 hours before she says Minc acknowledged responsibility and took it down.”

• 
  Amitoj Gautam (@amitojgautam) reported 59 minutes ago

  @airtelindia @Airtel_Presence I’m facing what appears to be a serious IPv6 routing/throughput issue on my Airtel broadband. My 300 Mbps connection gives ~300 Mbps download AND upload on Speedtest. However, with IPv6 enabled, Dropbox uploads collapse to around 10–15 KB/s, and services such as Gmail and some websites also become extremely slow/unresponsive. After disabling IPv6 on the Ethernet adapter, Dropbox immediately jumped to ~39 MB/s and the affected websites started loading normally within seconds. This has been reproduced consistently, so it does not appear to be a general bandwidth or Dropbox issue. Please escalate this to the network/IPv6 team and check IPv6 routing, packet loss, MTU/PMTUD and provisioning on my connection.