Mythos models demonstrated critical cybersecurity vulnerability detection capabilities
Mythos models previously found thousands of critical cybersecurity vulnerabilities, validating their security research capabilities.
Event / Technology
Claude Fable 5 AI Model Release
Anthropic released Claude Fable 5, a publicly available AI model with safety safeguards, alongside Claude Mythos 5 for trusted cybersecurity partners. The model excels at software engineering and reasoning tasks, though UK researchers have made early progress toward jailbreaking its safeguards.
3 reported locations
Daily activity
0 articles0 articles
2026-08-16 to 2026-11-13 · UTC publication dates
Coverage begins 10 Jun, 02:36 UTC
Follow the documented timeline or explore the related reporting.
What happened
Anthropic releases Claude Fable 5, a publicly available Mythos-class AI model
Anthropic released Claude Fable 5 on Tuesday, marking the first public release of a Mythos-class AI model. The company describes Fable 5 as its most capable model ever made generally available, demonstrating state-of-the-art performance on nearly all tested benchmarks in software engineering, knowledge work, vision, and scientific research. Fable 5 uses the same underlying technology as Anthropic's more restricted Mythos models but includes safety guardrails to limit access to dangerous capabilities.
President Trump signs executive order enabling voluntary AI safety testing by federal government
President Donald Trump signed an executive order that establishes a voluntary mechanism for AI companies to share their systems with the federal government for safety testing up to 30 days before public release. The order allows the government to strengthen its own cyber defenses but explicitly prohibits mandatory review of AI companies.
UK AI Safety Institute produced early jailbreak attempts against Fable 5 safeguards
The UK AI Safety Institute made early inroads toward developing a universal jailbreak of Claude Fable 5's safety measures during initial testing, though Anthropic characterizes this as an acceptable risk. The company conducted external bug bounty testing that ran over 1,000 hours without producing a universal jailbreak.
Anthropic implements safeguards that route sensitive queries from Fable 5 to older Opus 4.8 model
Anthropic deployed safety safeguards on Claude Fable 5 that intercept queries on high-risk topics including cybersecurity, biology, and chemistry, routing these requests to the less powerful Claude Opus 4.8 model instead. The company stated that fewer than five percent of user sessions trigger this fallback mechanism. Anthropic acknowledged that the conservative safeguards may occasionally block benign requests but prioritized safety at launch.
Anthropic prices Fable 5 at $10 per million input tokens and $50 per million output tokens
Anthropic set pricing for Claude Fable 5 at $10 per million input tokens and $50 per million output tokens, which is less than half the cost of the previous Claude Mythos Preview and twice the price of the standard version of Claude Opus 4.8.
Stripe completed 50-million-line Ruby codebase migration in one day using Fable 5
Fintech company Stripe, which had early access to Claude Fable 5, reported that the model completed a full migration of a 50-million-line Ruby codebase in a single day. Anthropic estimated that this task would normally require a full engineering team more than two months to complete.
Anthropic releases Claude Mythos 5 to trusted cybersecurity partners without safety restrictions
Anthropic released Claude Mythos 5, an unrestricted version of Fable 5, to a limited group of trusted cybersecurity professionals and infrastructure providers through its Project Glasswing program. The company stated that Mythos 5 has the strongest cybersecurity capabilities of any AI model in the world and plans to expand access through a broader trusted access program.
Anthropic filed confidential S-1 draft, officially beginning IPO process
Anthropic announced that it had confidentially filed a draft S-1 with the Securities and Exchange Commission, officially initiating the process toward a public initial public offering.
Anthropic's Mythos models previously found thousands of critical cybersecurity vulnerabilities
Anthropic's Mythos Preview model, which lacks safety safeguards, demonstrated the ability to find thousands of critical and severe cybersecurity vulnerabilities, including bugs and exploits in all major operating systems and web browsers. This capability raised concerns among AI researchers about the potential for powerful AI systems to be misused in cyberattacks against critical infrastructure.
IMC trading analysis evaluations show Fable 5 excels in financial reasoning tasks
Financial services firm IMC evaluated Claude Fable 5 and found it excelled across nearly all trading analysis evaluations, including factual lookup, conceptual reasoning, root-cause analysis, and expected-value analysis.
OpenAI confidentially files S-1 draft, beginning its own IPO process
OpenAI, Anthropic's chief rival, announced that it had confidentially filed its own S-1 draft, officially beginning its path toward a public initial public offering.
Claude Fable 5 available to all subscribers at no additional cost through June 22
Anthropic made Claude Fable 5 available to all users on its Pro, Max, Team, and seat-based Enterprise subscription plans at no additional cost until June 22. After that date, users will need to purchase additional computing credits to access the model. The company stated it aims to incorporate Fable 5 into standard subscription plans as quickly as possible.
Read the coverage
2 alternative perspectives
Trump administration enables voluntary federal AI safety testing framework
Trump administration creates voluntary federal AI safety testing framework for companies.
Anthropic and OpenAI both file for IPO as AI race intensifies
Anthropic and OpenAI both filed for IPO, marking a pivotal moment in AI companies' path to public markets.
UK researchers successfully demonstrate Fable 5 jailbreak vulnerabilities
UK AI Safety Institute successfully jailbroken Fable 5 safeguards during early security testing.
Anthropic releases Claude Fable 5 with tiered access model
Anthropic released Claude Fable 5 with tiered access, performance safeguards, and demonstrated capability in complex tasks.