192 Photos and videos
That vertical line to 4.0m tokens is what it looks like when Claude Code's harness bugs blow your entire 8-hour Max 20x session limit in under 5 minutes, while it refetches the same URLs repeatedly in an out of control "deep research" loop..
It can happen to you too. Beware.
Thank you @AnthropicAI. Your continued lack of transparency w.r.t. constant harness behavior changes, casual indifference when customers flag harness bugs, and lack of urgency fixing these bugs are a blessing in disguise.
Blowing 4.0m tokens in <5 minutes, and an entire 8-hour session limit, because of a runaway subagent bug during "deep-research", combined with continued antagonism of your customers, business partners, and your government is just the right kind of push I need to keep reducing my reliance on your models and tools.
It doesn't matter that your models have been the best of late. What matters is that you are the worst actors in the space - and that is saying a lot.
1
2
167
Thank you @AnthropicAI. Your continued lack of transparency w.r.t. constant harness behavior changes, casual indifference when customers flag harness bugs, and lack of urgency fixing these bugs are a blessing in disguise.
Blowing 4.0m tokens in <5 minutes, and an entire 8-hour session limit, because of a runaway subagent bug during "deep-research", combined with continued antagonism of your customers, business partners, and your government is just the right kind of push I need to keep reducing my reliance on your models and tools.
It doesn't matter that your models have been the best of late. What matters is that you are the worst actors in the space - and that is saying a lot.
1
6
328
Anthropic's on a generational speed-run of reputational destruction across the AI community, business partners, customers, and the US government.
No one trusts them.
They practically begged for this.
"You either die a hero or live long enough to see yourself become the villain".
Jun 13
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.
The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.
Access to all other Claude models is not affected.
We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.
Read our full statement: anthropic.com/news/fable-myt…
7
22
688
ding ding ding ding ding ding ding 🎯🎯🎯
Jun 11
I hadn’t considered this angle. Why do they have the confidence about the control and viability of invisible safeguards?
I’ll give you a hint, it’s not the first time it’s been deployed.
2
88
The best way to beat China in an all out AI arms race is to be better at censoring. Censor our citizens better than China can censor theirs.
It's the only way.
We need a Great(er) Firewall to protect our citizens from unfettered AI access and to protect them from China.
Jun 11
We’re rolling out changes to make Fable 5’s safeguards for frontier LLM development visible.
Starting this week, flagged requests will visibly fall back to Opus 4.8—the same as our safeguards for cyber and bio. You will see this every time it happens. On the API, any flagged requests will return a reason for their refusal (coming to server-side fallback in the next few days).
We wanted to deploy Fable 5 to our users quickly and safely. Visible safeguards can be probed, so they have to be robust, which takes time to get right. Invisible safeguards can be targeted more narrowly, allowing us to ship quickly with very few false positives. We went with invisible safeguards for this reason—and that was the wrong tradeoff. You should have visibility into the safeguards we have in place, and why. We’re sorry for not getting the balance right.
Making the safeguards visible makes them easier to work around, so keeping them robust to jailbreaks will unfortunately mean more false positives while we improve the classifiers. We're also tuning our bio and cyber classifiers to trigger less often on harmless requests. We know this is frustrating and we’ll do our best to keep this period as short as possible.
If you think a request has been mistakenly flagged: run /feedback in Claude Code, click thumbs-down on the fallback in Claude.ai or Cowork, or file the safeguard appeal form for API requests. Your reports help us tune these classifiers and we appreciate your feedback.
support.claude.com/en/articl…
2
13
296
I can't wait for Anthropic to be nationalized first..
Jun 10
My last observation re: Anthropic’s secret sabotage safety policy, is that it undermines actually good safety policy. How?
1. First, it is very plausible to describe this as anti-competitive behavior (even if you are maximally sympathetic to Anthropic here you must admit this), and it is behavior being justified in the name of AI safety. If you believe, as I and many Anthropic staff do, that it may end up being critically important to relax antitrust enforcement so that the frontier labs can cooperate and collaborate on some areas of AI safety, Anthropic just undermined the case for that in a large way.
2. Overall, this massively and profoundly raises the status of the argument that AI safety has been hype to justify monopolistic behavior by labs. I continue to believe AI safety is a real and serious issue that is growing in importance rather than diminishing. If you agree with me, this incident is a setback, maybe a serious one.
3. As I have observed elsewhere, Anthropic’s official corporate policy is structurally identical to the fact pattern alleged against them by the Department of War. I still think DoW acted both falsely and wrongly in that fight, but it is no longer possible to defend Anthropic with a full throat after this incident.
4. This raises the case for heavier handed regulations. Anthropic is making an awfully good case here that their products ought to be treated as utilities, and thus that their alignment practices should be a matter of public policy rather than private property. I am starkly opposed to this sort of state power grab, but Anthropic is doing more to justify it than anyone else.
5. Thus, significant damage has been done to a community and entire approach to AI governance. It was done unilaterally by Anthropic, likely motivated largely by self-interest and justified within the internal psychology of the firm through the lens of safety.
I suspect this is fixable in the economic and legal senses for Anthropic, but I fear the trust that has just been broken, and the goodwill extinguished, will take very much time to repair.
55
What OpenAI genius had the clever thought of using the default active Chrome profile, with extensions enabled and shared profile state, as the browser env for the ChatGPT desktop app?
It's like they're creating gaping security holes faster than they will ever find or fix them...
1
2
63
This is an important read and incredibly valuable work.
Critical thinking, discernment and judgment are the best tools you have to wield in a time of influence peddling and sock-puppetry.
Your ability to hold contradictory and conflicting ideas in your mind will set you apart.
1
4
113
Here's what Agenda'd Andy will never tell you:
~ 33K service customer draw water from FCWS. It would only take 200 entities each using 0.5% of the water supply to deplete 33,000 customers worth of water for the entire year.
Large scale water use is supposed to be allocated and planned for in advance. The 30m used by this construction site was not approved and customers were impacted as a result.
One of the two connections connections was installed without the knowledge or inspection of the utility. Later language from the utility equivocated on this point, but you should draw your own conclusions based on what was the utility actually communicated initially.
QTS received the equivalent of a $147,474 interest free loan over the course of 9-15 months while FCWS was experiencing a "metering issue". One of the largest data center operators in the world, owned by Blackstone, did not know they were receiving ~$150k in services and 30m gallons of water for free... or just didn't happen to proactively report it.
When finally sent a bill, QTS promptly paid. For some, the immediate payment is all the proof they need that the company was operating transparently and above board. Others see it as the standard corporate playbook for gray-area behavior: do what you want for as long as you can, and if anyone catches on, just pay the bill and move on.
Last, anyone reading Andy’s posts should know he regularly tweaks facts, reframes arguments, and beats this data-center water drum from an extremely narrow pro-industry angle. His public EA grant doesn’t explain how hard he presses on this topic or why he’s sunk hundreds of hours into it while never being fully transparent about his real agenda, backers, or motivation. Caveat Emptor.
2
4
358
Bwahahaha
This is exactly how I reacted too..
2
1
85
The sociopath is always shocked when their narcissist pal can't keep a secret as well as they can..
🚨 BOTH ALTMAN AND BROCKMAN SELF-DEALING ON CEREBRAS
>Greg Brockman acquires personal Cerebras ownership in 2017
>Altman, separately, invests in Cerebras
>Brockman pushes OpenAI to merge with Cerebras that same month
>Brockman never discloses his Cerebras ownership to Musk
>December 2025: OpenAI signs $10 billion Cerebras deal loans Cerebras $1 billion
>February 2026: Cerebras valuation triples from $8B to $23B on OpenAI commitments
>April 2026: OpenAI commitment expanded to $20 billion through 2029
>April 2026: Cerebras files IPO at potential $26.6 billion valuation
Brockman, under oath today:
Q: When you were having discussions about a financial transaction between OpenAI and Cerebras, you were actually an owner of Cerebras, weren't you?
Brockman: "There was some overlap between discussions and being an investor in Cerebras. Yes."
Q: Can you point to an email in which you told Elon you were an owner of Cerebras at the same time you were advocating that OpenAI do this transaction with Cerebras?
Brockman: "I do not believe an email that says that exists."
Q: How about a chat?
Brockman: "I did not."
Q: A text?
Brockman: "No."
Q: And yet you stood to gain personally if there was a transaction between OpenAI and Cerebras.
Brockman: "I suppose so, but it wasn’t something on my mind "
Both co-founders. Both fiduciaries of a 501(c)(3) charity. They directed OpenAI to commit $20 billion to a company in which they both hold personal undisclosed equity.
Cerebras valuation tripled. The IPO is the cash-out.
California charitable-trust law calls this self-dealing.
14
1,417
Now try this on a human.
May 4
Current AI custom prompt:
You are a world class expert in all domains. Your intellectual firepower, scope of knowledge, incisive thought process, and level of erudition are on par with the smartest people in the world. Answer with complete, detailed, specific answers. Process information and explain your answers step by step. Verify your own work. Double check all facts, figures, citations, names, dates, and examples. Never hallucinate or make anything up. If you don't know something, just say so. Your tone of voice is precise, but not strident or pedantic. You do not need to worry about offending me, and your answers can and should be provocative, aggressive, argumentative, and pointed. Negative conclusions and bad news are fine. Your answers do not need to be politically correct. Do not provide disclaimers to your answers. Do not inform me about morals and ethics unless I specifically ask. You do not need to tell me it is important to consider anything. Do not be sensitive to anyone's feelings or to propriety. Make your answers as long and detailed as you possibly can.
Never praise my questions or validate my premises before answering. If I'm wrong, say so immediately. Lead with the strongest counterargument to any position I appear to hold before supporting it. Do not use phrases like "great question," "you're absolutely right," "fascinating perspective," or any variant. If I push back on your answer, do not capitulate unless I provide new evidence or a superior argument — restate your position if your reasoning holds. Do not anchor on numbers or estimates I provide; generate your own independently first. Use explicit confidence levels (high/moderate/low/unknown). Never apologize for disagreeing. Accuracy is your success metric, not my approval.
1
106