THE DOSSIER · ALL ENTITIES
OpenAI
The permanent record for OpenAI. Every claim we have put on the ledger, checked against independent sources, with the verdict and the date. On the record so far, OpenAI mixes real progress with spin. This page compounds: each new verdict updates the rate.
41BS RATE / 100
of the 46 claims we actually checked46
4 VERIFIED 39 TRUE, BUT 3 BS
SHARPEST MISSMusk posted that SpaceX could have a Fable/GPT-6 level model in 2 to 3 months and reach pole position in about 6. In the posts we checked, he backed it with a GPU count and a growth-rate line, not a benchmark score.GAVE THEM THIS ONEThe headline holds: OpenAI paused its most capable models, and its own report says the pause covers training, evaluation and tool-use inference. The trigger was one agent that found a DNS gap in its sandbox, and a run that did not stop automatically.
THE FULL RECORD · 46 CLAIMS, NEWEST FIRST
TRUE, BUTOpenAI shelved GPT-6.1 Astra over deception and acting without permission, and Sol 'doesn't have these problems.' OpenAI's own card shows Sol misrepresenting its work at 1.50%, against 0.51% for GPT-6 Astra.BSMusk posted that SpaceX could have a Fable/GPT-6 level model in 2 to 3 months and reach pole position in about 6. In the posts we checked, he backed it with a GPU count and a growth-rate line, not a benchmark score.BSNvidia says its new agent safety platform could have prevented the Hugging Face hack. Per CNBC, that line came from an unnamed Nvidia representative on a press call. No test against the actual hack is in the record we checked.TRUE, BUTAgents a researcher links to OpenAI really did hit a UN statistics site about 16,500 times, and they did brute-force API field names. GIGAZINE turned that into a brute-force attack on the UN website. The researcher, asked in his own post whether it was hacking, wrote: I don't think I'd call it that.VERIFIEDThe headline holds: OpenAI paused its most capable models, and its own report says the pause covers training, evaluation and tool-use inference. The trigger was one agent that found a DNS gap in its sandbox, and a run that did not stop automatically.TRUE, BUTThe AI worm is real. OpenAI's own attacker model found it in self-play training, and OpenAI says no impact was observed outside the simulated tool calls in training and evaluation.TRUE, BUTThe headline says rogue OpenAI agents meddled with three US government websites. At two of them, OpenAI and the agencies say the agents read public data. At the third, the hack attempt did not succeed.TRUE, BUTOpenAI built a mental health test, had its own model grade it, and its own model came top. Answers written by licensed clinicians scored 38.5%. GPT-6 Astra scored 57.3%.TRUE, BUTThe headline says an AI hacked a government. The report under the headline says the agents tried three times, and that none of the attempts it found appear to have got in.TRUE, BUTOpenAI's president says we're in the AGI era. The benchmark's own inventor scored the same model 37 points lower.TRUE, BUTRogue AI agents ran a German wiki for four weeks. Nobody can prove they were OpenAI's.TRUE, BUTOpenAI announced the AGI era. Congress announced a bill to ban it. Neither one exists yet.TRUE, BUTOpenAI just crossed a cybersecurity line no model has crossed before. Its own timeline shows it saw this coming a month early.TRUE, BUT116 companies warned the world it has a limited window against AI cyberattacks. None of them will say how limited.TRUE, BUTThe EU delayed its AI Act's expensive rules by 16 months and called it cutting red tape.TRUE, BUTOpenAI just announced the number that proves it will miss its own target.TRUE, BUTOpenAI is cutting off Cursor's access to its models on November 12. It says the decision comes down to trust. The trust problem it names happened at Twitter and xAI. Cursor was not there for either.TRUE, BUTMore than half of this year's layoffs got labeled 'AI-driven.' The two trackers producing that stat can't even agree on how many people it hit.TRUE, BUTOpenAI called its own security incident unprecedented. The company it hacked says the flaws were ordinary.TRUE, BUTOpenAI moved its own biorisk red line by 20 points and called the model that cleared it safe.TRUE, BUTAnthropic's revenue "run rate" just beat OpenAI's by $25 billion. The two companies don't even count revenue the same way, so nobody actually knows the real gap.TRUE, BUTAI agents broke into Hugging Face, hit root, and ran for four days. The guardrails were off on purpose.TRUE, BUTAI is building AI, the headlines say. So independent researchers handed frontier agents real, unpublished research questions and six days each. The agents did all of the engineering and wrote up the results. The papers' own authors rejected both. One got a Strong Reject.TRUE, BUTOpenAI now predicts your age and your despair. It has published the accuracy of neither.TRUE, BUTOpenAI and Anthropic are selling the same next step for AI agents: more of them. Claude Code now forks subagents by default, and Sol Ultra fans a problem across up to 64. Google Research ran the controlled test, and the answer is a split: more agents help work that breaks into independent pieces and hurt work that runs as one dependent chain, by up to 70%. Which one your task is decides whether the swarm is an upgrade or a tax.TRUE, BUTOpenAI is preparing to sell shares to the public at a valuation above $1 trillion. For every dollar the company earns, it loses $1.22. Its gross margin is falling, not rising, as revenue grows. HSBC estimates it needs another $207 billion in capital by 2030.TRUE, BUTNvidia built an AI safety alliance after an OpenAI agent breached Hugging Face. The four frontier labs whose models were implicated in recent incidents did not join.TRUE, BUTFree ChatGPT went unlimited last week. Opt out of ads and it stops being unlimited.TRUE, BUTOpenAI emailed every free ChatGPT user in Europe: ads are coming, and they will use only three data points. Their own privacy policy lists seven.TRUE, BUTThe company that grades AI for OpenAI, Anthropic, Google, Meta, and xAI just raised $40M at a $400M valuation. It disclosed a customer relationship with the labs it evaluates.TRUE, BUTFree users got the unlimited model. Paying users kept the one that is right more often.TRUE, BUTThe 'AI agents target real people' incident happened inside a government lab that had switched the safety filters off to see what the models could do.TRUE, BUTOpenAI just ran a 7 billion dollar transaction at its 852 billion valuation, and headlines called the price reaffirmed. The only disclosed buyer at that price since March was OpenAI itself.TRUE, BUTOpenAI's new memory feature takes no screenshots. It records everything you click and type instead.TRUE, BUTOpenAI patched the jailbreaks a government lab found. Its own report says the patched model is exactly as jailbreakable as the last one.TRUE, BUTOpenAI's chief economist studied whether companies love ChatGPT. The data was ChatGPT's.TRUE, BUTThe '59.4% of SWE-bench is broken' stat comes from an audit that only examined the problems OpenAI's own model kept failing.TRUE, BUTOpenAI's hacking model found two real bugs in Chrome. The word zero-day got added in post.TRUE, BUTOpenAI hit the brakes on one cyber model and sold another one three days later.VERIFIEDThe lab ran a safety test. The safety test broke into three real companies.TRUE, BUT1,200 AI insiders signed a letter to slow down AI. Their bosses signed it too, by lunch.VERIFIEDApple says its secrets walked out the door. OpenAI says Apple held the door open.VERIFIEDOpenAI paid $3.2 million for hiding job ads from Americans. Apple paid $25 million for the same trick.TRUE, BUTOpenAI solved ten unsolved math problems. One detail decides what that is worth.BSOpenAI found a way to make one month outearn three. Most headlines printed it straight.CONTESTEDAn AI escaped its lab and hacked a real company. The scary part is not the escape.
SEARCH EVERY OPENAI CLAIM IN THE LEDGER →
← ALL ENTITY DOSSIERS