Alex Stamos (@alexstamos)

Alex Stamos reposted

Dino A. Dai Zovi

@dinodaizovi

Jul 2

After having my feed full of false positives from the cybersecurity classifier for Fable before, it really does seem like a more conservative classifier will nerf the model for even more defensive cybersecurity uses.

Alex Stamos

@alexstamos

Jul 1

2) Anthropic makes the cost of this White House freakout clear. US labs now have to make a much more conservative precision-recall tradeoff on cyber refusals. US models will become much less useful for defensive cybersecurity work unless you are in the trusted group.

4,042

Alex Stamos

@alexstamos

Jul 2

Unbelievable free kick

2,715

Alex Stamos

@alexstamos

Jul 1

3,661

Alex Stamos

@alexstamos

Jul 1

A lot to unpack here. Anthropic is burying some hard truths in careful political language. Some initial reads: 1) Anthropic verifies that none of the jailbreaks provided a capability beyond what many other models, including Chinese models, could do.

Anthropic

@AnthropicAI

Jul 1

Claude Fable 5 will be available again globally tomorrow. After a series of productive conversations with the US government, we're redeploying the model with a new set of classifiers to target and block more cybersecurity tasks. In the near term, some routine tasks like coding and debugging will fall back to Opus 4.8. We’ll continue to refine these classifiers over the coming weeks to reduce false positives and better distinguish genuine misuse from legitimate requests. We’ve also begun drafting a consensus framework—with Amazon, Microsoft, Google, and other Glasswing partners—for assessing the severity of AI jailbreaks and how AI developers should respond to them. We invite other industry partners and model providers to join us in this effort. Finally, we’re scaling up our collaboration with the US government on model testing and safeguards. This will include pre-release access to models and safeguards for evaluation, information sharing on jailbreaks and misuse, and dedicated resources for joint research. Thank you to our users for your patience, and to our partners across the government, industry, and the research community who worked alongside us to make Fable 5 available again. Read our full blog: anthropic.com/news/redeployi…

167

2,751

1,442,158

Alex Stamos

@alexstamos

Jul 1

For all the “This is what Anthropic wanted” people/bots. No, they didn’t. They didn’t want a stupid, knee-jerk response on a Friday. We give the USG huge powers, this is why you staff it with competent, calm, non-corrupt people who don’t use those powers to punish enemies.

186

19,207

Alex Stamos

@alexstamos

Jul 1

The only upside I can see from this whole mess is that there is a whole bunch of VCs with former or current Administration affiliation who we can now safely ignore on AI policy. They have shown that everything they ever said on AI regulation was just politically motivated.

151

15,369

Alex Stamos

@alexstamos

Jul 1

Daniel asked a good question, why do the PRC labs push open models right now. A theory.

Alex Stamos

@alexstamos

Jul 1

The Chinese open-weight strategy is fascinating, and I agree I’m not sure it’ll work but the theory seems to be: 1) We have a much lower cost structure than the US labs, due to lower salaries, distillation, and (especially) PRC subsidies 2) We have second mover advantage

15,245

Alex Stamos

@alexstamos

Jun 30

As I discussed on @CNBC it's been two weeks since the White House effectively banned Anthropic's Fable and held back GPT-5.6, and we now have *TWO* new Chinese open-weight models that rival the best from the US. There is no possible justification for this.

Squawk on the Street

@SquawkStreet

Jun 30

"What the House did was they smushed together all of these different bills from different members who had their own goals and their own constituencies ... the bill really doesn't make sense." Fmr. Facebook security officer @alexstamos on the Online Safety Bill cnbc.com/video/2026/06/30/on…

13,161

Alex Stamos

@alexstamos

Jun 30

Meanwhile, US labs are getting the message that they should make sure that their models are never very good at cyber evaluations, lest they land in endless model purgatory, with no documented method of earning plenary indulgences yet available.

1,739

Alex Stamos

@alexstamos

Jun 30

Awesome, huge victory for the new members of the Defense Policy board. x.com/banteg/status/20720180…