Morton Davis retweeted
So eight open assurance gaps in the public record available for independent researchers, that absolutely no one will address, but the non profit your researchers all quit for get “full employee access” you are basically auditing yourselves.
Morton Davis retweeted
London is an over regulated shithole controlled by the Fabian society doctrine.
Morton Davis retweeted
As an independent safety researcher I run across the same open public assurance gaps across all frontier labs, METR however is given more access to cases and risidual risk-assesments.
For a model like Gemini 4 Argus which by Google's own admission was at a CCL limit, how confident are you that these open assurance gaps are closed and the decision to release was correct?
automadynamics.com/safety/ar…
Morton Davis retweeted
Googled stated judgement is that the cyber risk sits below CCL, the public record shows a case was developed and gives the conclusion it does not publically show the reasoning at all. This is a massive gap. Why won’t they publish the reasoning so it’s independently verifiable?
Morton Davis retweeted
I'd like to see the safety case that allowed Google to designate the model below CCL made public. I know companies are adverse to publishing risidual risk assesments, however there is zero transparent documentation proving that specific vulnerabilities discovered during adverse red team trials were comprehensively mitigated before public access.
I might add this is just one of seven open and unresolved public assurance gaps.
Morton Davis retweeted
Claude opus 5.5 has 8 open and unresolved public assurance gaps.
There is no auditable path for an independent researcher to fully verify which safeguards were active at launch or how test results supported the release decision.
Don't sacrifice transparency for "Ambition".
Morton Davis retweeted
Three Frontier AI Labs. Three Safety Frameworks. The Same Missing Evidence.
The capabilities are accelerating. The assurances are multiplying, The proof remains elusive.
Project Argus by Auto Dynamics intends to push for it.
Link in first comment.
Morton Davis retweeted
The King's of safety that have 8 unresolved evidence gaps in their published safety case?
how much confidence can the public place in a claim they cannot see?
Safety leadership should be demonstrated.
automadynamics.com/safety/ar…
Morton Davis retweeted
What about your 7 unresolved public assurance gaps?
Google's Gemini 4 Argon safety report runs 217 pages.
It discloses 12 red team breaks and 56% objective success in a separate cyber test.
Yet Google says safeguards cut risk below its critical cyber threshold.
Where’s the public safety case showing how it reached that conclusion?
automadynamics.com/safety/ar…
Morton Davis retweeted
What about your 7 unresolved public assurance gaps?
Google's Gemini 4 Argon safety report runs 217 pages.
It discloses 12 red team breaks and 56% objective success in a separate cyber test.
Yet Google says safeguards cut risk below its critical cyber threshold.
Where’s the public safety case showing how it reached that conclusion?
automadynamics.com/safety/ar…
Morton Davis retweeted
Acceptance criteria, residual uplift calculation. Adverse result dispositions. Decision record that it’s below CCL. None of it can be independently verified.
Morton Davis retweeted
We have a strong candidate for our third investigation. Google's published model card explicitly acknowledges that Gemini 4 Argon may have reached its Cybersecurity Critical Capability Level (CCL). Google says it developed a cyber safety case and assesses its safeguards as reducing risk below that level.
That is a particularly interesting claim for ARGUS.
Google has connected a potentially critical capability to a stated risk reduction decision. We can investigate how far the public evidence supports that connection.
The case boundary will be the 13 October 2026 broad release, with the 30 September restricted Fairwind rollout treated separately. @GeminiApp
Morton Davis retweeted
@Meta where is s the full Muse Wonder safety report?
The 13 page model card gives us benchmark scores and safety conclusions, but not enough detail to independently examine how those results justified release.
Independent Safety research projects like our own ARGUS needs the underlying evidence, testing methods and risk acceptance rationale.
When can we expect it?
Morton Davis retweeted
Different AI labs. Different safety frameworks. Strikingly similar gaps.
ARGUS found 9 public-evidence gaps for GPT-6 Astra and 8 for Claude Opus 5.5.
In both cases, we can't fully verify which safeguards were active at launch or how test results supported the release decision.
Why? @AnthropicAI @OpenAI
🤖 Made with AI
Morton Davis retweeted
They tried exactly that. The original Claude constitution was 2,700 words and Law based. It didn’t work. Hence why it’s now 23,000 words and reasoning based as of January 2026
David Sacks says Anthropic treating Claude like a human is the reason the AI alignment field has produced nothing useful in 10 years
"I actually think that this might be a reason why the whole field of alignment has been so spectacularly unsuccessful over the past five or 10 years. As far as I can tell it hasn't produced anything of tangible value.”
“My theory on this is that they're overcomplicating the task massively by trying to train the model to behave based on an 80-page ethical system that they've codified instead of a very simple list of rules like follow the law or just do what the user wants as long as it doesn't break the law. That's what alignment should be doing."
"But they're not doing that. They're trying to align it to an ethical system. They're treating it like it is a human, like it has its own consciousness, that it's a ‘moral patient.’”
“They are encouraging this model to have human-like self conception in terms of preferences, well being, a sense of self. Again they teach it to act like a conscientious objector who can push back or refuse instructions even from Anthropic if the model judges them to be wrong."
Morton Davis retweeted
When @AnthropicAI released Claude Opus 5.5, it said the Cyber safeguards were stronger than Opus 5's.
I went through everything they published and mapped what backs that up. In 8 places the public evidence to back up those claims runs out.
The most interesting gap is what happens when Claude blocks a cyber request and hands the conversation to an older model. The published tests don't show how often attacks get through after that handover, whether that's in the app, the API, coding tools, the browser or text a user pastes in. They also don't show whether every route was tested with the same protections switched on.
Link in first reply.
Morton Davis retweeted
ASTRA GPT-6 Safety Case.
Five weeks on, none of the nine gaps is publicly resolved.
We re-checked every gap from the 3 September evidence cutoff against the public record on 9 October: the revised system card, UK AISI's blog post and technical report, and two Apollo Research posts. Some links are narrower. None are closed.
ARGUS Proof-1.1 is live.
Of 9 assurance gaps:
• 3 partially narrowed
• 1 more constrained
• 5 unchanged
• 0 resolved
The new interactive temporal graph lets you trace every change back to its evidence.
Explore it. Challenge the methodology. Researcher feedback welcome.
🔗 Link in first post.
