UK AI Safety Institute confirms Claude Mythos finds more exploits as token spend increases
The UK's AI Safety Institute published an independent evaluation confirming Anthropic's Claude Mythos is highly effective at finding security vulnerabilities. The evaluation revealed a linear relationship: more tokens spent equals more exploits discovered, transforming security into an economic arms race.
UK AI Safety Institute confirms Claude Mythos finds more exploits as token spend increases
The UK's AI Safety Institute (AISI) published an independent evaluation of Anthropic's Claude Mythos Preview, confirming the model's capabilities in identifying security vulnerabilities. The evaluation, titled "Our evaluation of Claude Mythos Preview's cyber capabilities," validates Anthropic's claims about the model's security testing effectiveness.
Token spend correlates with vulnerability discovery
The AISI report reveals a critical finding: Claude Mythos continues discovering exploits proportionally to token expenditure. According to analyst Drew Breunig, this creates a straightforward economic equation for cybersecurity: "to harden a system you need to spend more tokens discovering exploits than attackers will spend exploiting them."
This transforms security testing from a qualitative practice into a quantitative proof-of-work problem, where defensive spending must exceed offensive spending to maintain system integrity.
Open source economics shift
The findings have significant implications for open source software development. Because token investments in securing open source libraries can be amortized across all users of those libraries, shared security costs make open source projects economically more attractive.
This counters recent arguments that AI-powered "vibe-coding" — rapidly generating custom code replacements — would diminish the value of established open source libraries. The security economics now favor reusing well-tested shared libraries over custom implementations.
What this means
Cybersecurity now has a measurable cost floor: the token expenditure required to match potential attacker budgets. Organizations must budget not just for security tools, but for the computational cost of thorough AI-assisted vulnerability discovery. This creates a stark divide between well-funded projects that can afford extensive security testing and under-resourced projects that cannot.
The shift also strengthens the economic case for open source infrastructure, as communities can pool resources for security audits rather than each organization bearing the full cost independently. Security becomes a shared computational expense rather than duplicated effort.
Related Articles
Anthropic Warns Small AI Models Are Months From Self-Replicating Botnets; Experts Call Threat Familiar
Anthropic's Frontier Red Team published research suggesting small AI models may be only months away from replicating across insecure devices as autonomous botnets. Experts told Engadget the capability is plausible, but that the framing overstates how new the threat is.
METR report: Anthropic's unreleased Model 4 scores 178.6 on AECI, ahead of Opus 5.5
METR's new report discloses three unreleased Anthropic models, Model 2, Model 3 and Model 4. Model 4 scores 178.6 on Anthropic's AECI index, about 3.9 points above Opus 5.5. Anthropic co-founder Sam McCandlish confirmed the report's main findings but said much information was removed.
METR says Anthropic redacted most of the evidence it found most informative in its ~100-page report
METR says it submitted a roughly 100-page full report on its assessment to Anthropic, and that Anthropic redacted a majority of the evidence METR found most informative from the public version. METR says readers must therefore rely on its judgment more than it considers ideal.
AI Running Costs Fell This Year Even as Server Prices Rose, but Data Center Curbs Test the Trend
The cost of running AI has fallen this year even as server rental prices rose. After U.S. voters elected officials who want to restrict data center expansion, The Information reports that some executives and investors worry capacity delays could stall that decline.
Comments
Loading...