Anthropic rolls out public version of Mythos without cybersecurity capability
Anthropic is rolling out a public version of its Mythos AI model, but with guardrails barring its use in โrisky areas such as cybersecurity, after an earlier preview this year sent shockwaves globally for its ability to find software flaws. The new Claude Fable 5 is the most powerful model Anthropic has ever made for wider use, the startup said on โTuesday, touting its performance in software engineering and analytics. Anthropic has so far โlimited its access to a group of about 200 organisations, including โ the US government under the Glasswing program, after announcing in April that โMythos had uncovered thousands of software vulnerabilities. Offering its capabilities more widely may allow the $965 โbillion company to extend the momentum that has powered its valuation above rival OpenAI just as the two startups at the centre of the AI industry race to go public. The โcompany said it had done extensive testing to ensure that users could not โmanipulate the new model to bypass its guidelines and perform restricted actions. โLetโs say Iโm a โcollege โ student asking the model like help me find cyber vulnerabilities on X package or code. The model would refuse and Fable 5 will fall back to Opus 4.8 for a response,โ Dianne Penn, Anthropicโs head of product management, research and โlabs, told Reuters. Fable โ5 will be โ a more expensive model, but it accomplishes tasks with lower token usage, bringing the overall cost per task down, according โto early customer feedback, Penn said. Anthropic also said users who โhad access โ to the preview version of Claude Mythos, the version without guardrails, would be able to upgrade to the new Claude Mythos 5. The company said it planned to โ expand access โover time through a more โsystematic trusted-access programmeโ. Pricing on โboth models is $10 per million input tokens and $50 per million output tokens, the company said.
