Anthropic plans wider release of 'Mythos' AI, launches Opus 4.8 for coding
Mythos AI can hack every major OS and browser, now deemed safe enough for public release.
Anthropic has revealed plans to release AI models with cybersecurity capabilities comparable to its Mythos technology, which the company previously deemed too dangerous for general public access. Mythos is capable of identifying and exploiting vulnerabilities in every major operating system and web browser. To mitigate risks, Anthropic initially limited access through Project Glasswing to a select group of large tech companies and Wall Street institutions earlier this year. Now, citing 'swift progress' in safety safeguards, the company intends to make similar models available to a broader customer base, though it did not provide a specific timeline.
Alongside this announcement, Anthropic launched Opus 4.8, a model designed to excel at coding tasks. This move comes as the company intensifies its competition with OpenAI, with strong momentum in AI coding offerings and consumer traction despite a standoff with the Pentagon over safety protocols. Anthropic is nearing the close of a funding round at a valuation exceeding $900 billion, surpassing OpenAI's latest valuation. Both companies are reportedly considering public offerings as early as this fall.
- Mythos AI can identify and exploit vulnerabilities in every major OS and web browser.
- Opus 4.8 is a new model optimized for coding tasks on behalf of users.
- Anthropic nears $900B valuation, planning to go public alongside OpenAI.
Why It Matters
Democratizing such powerful cybersecurity AI could reshape enterprise defense but raises critical safety and ethical questions.