Anthropic is currently testing a new artificial intelligence model referred to internally as Claude Mythos. The company describes the system as a substantial leap forward in capability compared to its existing models. The development has drawn attention following the emergence of leaked internal documents that raise serious concerns about the technology’s potential risks.
According to those leaked documents, Claude Mythos could meaningfully increase cybersecurity threats by rapidly identifying and exploiting vulnerabilities in software systems. Internal warnings suggest the model’s capabilities in this area are advanced enough to potentially accelerate what observers describe as a cyber arms race. The documents indicate Anthropic is aware of these risks as it continues its testing process.
The news has had a visible effect on financial markets. The iShares Expanded Tech-Software Sector ETF, traded under the ticker IGV, fell by nearly 3% in early Friday trading. Analysts suggest this decline is likely connected to broader market unease surrounding the revelations about the new model.
The drop in IGV is also seen as a contributing factor to a retreat in cryptocurrency markets. Bitcoin slipped back to approximately $66,000 during the same trading period, reflecting wider risk-off sentiment among investors. The simultaneous moves across tech equities and digital assets point to the broad market sensitivity surrounding developments in advanced AI.
The situation highlights ongoing tensions within the artificial intelligence industry over how to balance rapid capability development with safety and security considerations. Leaked internal materials from AI companies have previously sparked public debate about the pace at which powerful systems are being developed and deployed. Anthropic has not yet made a public statement addressing the leaked documents or the specific concerns they raise about Claude Mythos.
Originally reported by CoinDesk.
