Blab Tech
Anthropic Unveils Fable 5, First Public Mythos-Class AI Model
When the afternoon sun slipped behind San Francisco’s skyline, Anthropic rolled out a new AI that could rewrite the rules of safety. On Tuesday, the company announced the public release of Claude Fable 5, its first Mythos‑class model made available to general users. The announcement came just a week after the U.S. president signed an executive order that encourages AI firms to share their systems with the government for safety testing.
Fable 5 is described by Anthropic as a Mythos‑class model that has been "baked in safeguards" to prevent misuse. It builds on the same underlying technology that powered the Mythos Preview system, which earlier uncovered thousands of critical vulnerabilities in operating systems and browsers. To curb potential abuse, Anthropic has wrapped the model in guardrails that block many requests about cybersecurity, biology, and chemistry. When a user submits a query that falls under these restricted categories, the system automatically redirects the request to the older Opus 4.8 model—less capable but still useful for general tasks.
The new model will be free for subscribers on Anthropic’s Pro, Max, Team, and Enterprise plans until June 22. After that date, users will need to purchase additional computing credits. Anthropic has said it intends to roll Fable 5 into all subscription tiers "as quickly as we can," but the company notes that it has struggled to keep pace with demand for its AI services.
In parallel, Anthropic announced that organizations that previously had access to Mythos Preview will now receive an upgraded Mythos 5 model. The preview version was offered to more than 150 companies—including banks, software firms, and healthcare networks—to help them strengthen cyber defenses before a wider public release. Mythos 5 has reportedly shattered performance records in drug design and molecular biology, with internal tests showing it can produce novel scientific hypotheses faster than human experts.
The release of Fable 5 and Mythos 5 comes amid heightened scrutiny of powerful AI systems. According to the company, the same capabilities that aid cybersecurity professionals can also be used by malicious actors to target critical infrastructure or develop bioweapons. Anthropic’s decision to deploy broad restrictions on sensitive topics reflects a "deliberately conservative approach" at launch, the company said in response to a question from NBC News.
The U.S. executive order, signed by President Donald Trump, establishes a voluntary mechanism for AI companies to share their systems with federal agencies for safety testing before public release. The order is intended to strengthen the government’s cyber defenses and to ensure that advanced AI models are vetted for potential misuse.
At present, Fable 5 is live for paid subscribers, and Mythos 5 is available to a limited set of trusted partners. Anthropic has not yet announced a public release date for the full Mythos 5 model. The company’s next steps will likely involve monitoring usage patterns, refining guardrails, and coordinating with government agencies under the new executive order. Unresolved questions remain about how effectively the redirect to Opus 4.8 will prevent misuse and how the broader AI community will respond to Anthropic’s approach to safety.
Fable 5 is described by Anthropic as a Mythos‑class model that has been "baked in safeguards" to prevent misuse. It builds on the same underlying technology that powered the Mythos Preview system, which earlier uncovered thousands of critical vulnerabilities in operating systems and browsers. To curb potential abuse, Anthropic has wrapped the model in guardrails that block many requests about cybersecurity, biology, and chemistry. When a user submits a query that falls under these restricted categories, the system automatically redirects the request to the older Opus 4.8 model—less capable but still useful for general tasks.
The new model will be free for subscribers on Anthropic’s Pro, Max, Team, and Enterprise plans until June 22. After that date, users will need to purchase additional computing credits. Anthropic has said it intends to roll Fable 5 into all subscription tiers "as quickly as we can," but the company notes that it has struggled to keep pace with demand for its AI services.
In parallel, Anthropic announced that organizations that previously had access to Mythos Preview will now receive an upgraded Mythos 5 model. The preview version was offered to more than 150 companies—including banks, software firms, and healthcare networks—to help them strengthen cyber defenses before a wider public release. Mythos 5 has reportedly shattered performance records in drug design and molecular biology, with internal tests showing it can produce novel scientific hypotheses faster than human experts.
The release of Fable 5 and Mythos 5 comes amid heightened scrutiny of powerful AI systems. According to the company, the same capabilities that aid cybersecurity professionals can also be used by malicious actors to target critical infrastructure or develop bioweapons. Anthropic’s decision to deploy broad restrictions on sensitive topics reflects a "deliberately conservative approach" at launch, the company said in response to a question from NBC News.
The U.S. executive order, signed by President Donald Trump, establishes a voluntary mechanism for AI companies to share their systems with federal agencies for safety testing before public release. The order is intended to strengthen the government’s cyber defenses and to ensure that advanced AI models are vetted for potential misuse.
At present, Fable 5 is live for paid subscribers, and Mythos 5 is available to a limited set of trusted partners. Anthropic has not yet announced a public release date for the full Mythos 5 model. The company’s next steps will likely involve monitoring usage patterns, refining guardrails, and coordinating with government agencies under the new executive order. Unresolved questions remain about how effectively the redirect to Opus 4.8 will prevent misuse and how the broader AI community will respond to Anthropic’s approach to safety.
