I am pleased to see that OpenAI’s new model is roughly on par with Gemini Flash…
- Source
- x.com
- Date
I am pleased to see that OpenAI’s new model is roughly on par with Gemini Flash and Opus 4.8 on prompt injection risk. Nice work! Evaluating and naming other labs turns out to be a great way to encourage them to train more aligned models. We will continue to do this until other labs pay more attention to safety. This is good for everyone and there is a lot of room left to go! We solved prompt injection in practice for Claude models about two months ago. But prompt injection is a significant security risk no matter what model you use, and it is important that the industry similarly spends more effort to train their models to be resistant to prompt injection, among other elements of model alignment. As models become more capable and central to businesses and economies, the risks only increase. We should be taking them seriously, and doing the right thing for our customers and the world.
Context
Boris Cherny, who built Claude Code at Anthropic, posted that OpenAI's newest model now performs roughly on par with Google's Gemini Flash and Anthropic's own Opus 4.8 on resisting . Prompt injection is when an attacker hides instructions inside content a model reads, a webpage, email, or file, hoping the model follows those hidden instructions instead of the user's actual request. It is a live security concern for any that browses the web or runs tools on a user's behalf.
Cherny also says Anthropic solved prompt injection in practice for Claude models about two months earlier. Both claims come from an Anthropic employee comparing his own company's models favorably against competitors, without naming the specific OpenAI model, publishing a test methodology, or citing scores. That makes this an attributed claim from a competing lab, not an independently verified result, even though the underlying security concern it describes is real and shared across the industry.
- prompt injection — An attack that hides instructions in content an AI will read — a webpage, email, or document — tricking it into following the attacker instead of the user.
- AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
- benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.
- alignment — The work of making AI systems actually pursue what their builders and users intend, rather than something subtly or dangerously different.
postAnthropic's Boris Cherny Signals Claude Chat and Cowork Are Merging
postClaude Tag moves to Fable 5.1 inside Slack for Team and Enterprise plans
postAnthropic previews Function Hooks, an unshipped Claude Code extension point- postAnthropic confirms zero data retention for Mythos-class models this fall
Checking sign-in…
Loading comments…