Fact check: Claude Opus 5.5 reverse-engineering "nerf" is a user report, not a confirmed change

Anthropic documents that Claude Opus 5.5 ships with conservative cyber safeguards. Its launch post says most cybersecurity tasks are re-routed to Opus 4.8, and a Help Center article says the checks also read web results and files, so content you did not type can trigger a stop. Anthropic's October 6 post calls the safeguards intentionally broad and points verified defenders to its Cyber Verification Program.

Users say this now hits the viral Reverse Engineer Anything (REA) tool, which 404 Media covered on October 20. A Hacker News commenter pasted the refusal: Opus 5.5 "is more capable and has stronger safeguards as a result, which can sometimes flag non-cybersecurity work". The same wording appears in a Claude Code bug report. Two small X accounts, @Octoberry and @Im_IrushiK, describe Claude stopping mid-answer on REA. They are user reports, not authoritative.

Not confirmed: that anything changed on October 21, that Claude was weakened on purpose, or that Anthropic targeted REA. Anthropic has said nothing about REA, and we could not open the Reddit threads, so we rely on the posts above. Kotaku is one outlet that covered the claim.

Sources: Anthropic, Hacker News, GitHub, X, 404 Media, Kotaku.

Why it matters

If your work involves binary analysis, expect some safety stops on Opus 5.5 and plan a fallback, or apply to Anthropic's Cyber Verification Program.

Source: Anthropic