This Is Bad... They Just Shut Down FABLE 5

The US government ordered Anthropic to abruptly shut down its Fable 5 and Mythos 5 models over vague national security concerns, locking out even Anthropic's own foreign-national employees just days after the launch. Anthropic argues the action fails any transparent, fact-based statutory test, noting the alleged jailbreak was minor and already achievable with other models like GPT-5.5 despite extensive pre-launch red-teaming. The move sparked backlash for undermining US competitiveness, pushing users toward Chinese models, even as Anthropic's momentum surges toward a trillion-dollar valuation and IPO. Public trust in AI companies is extremely low, with broad bipartisan support for government oversight, but this shutdown could set a dangerous precedent: if a narrow jailbreak justifies pulling a commercial frontier model, almost every model is at risk, creating a capability-access gap where powerful AI exists but the public is denied it.

• ▶ 0:02 The US government ordered Anthropic to shut down its Fable 5 and Mythos 5 models for all customers, citing national security concerns under export control authorities.
• ▶ 0:45 The shutdown was extremely broad: even Anthropic’s own foreign national employees were locked out of the models, forcing a complete plug-pull while regular Claude kept working.
• ▶ 1:18 Context: Fable 5 launched just three days earlier, bringing Mythos-like capabilities to the public, and its major capability jump was shown by beating Pokémon Fire Red in testing.
• ▶ 1:58 The government's letter lacked specific national security concerns; officials were allegedly shown a jailbreak for Fable 5, but Anthropic says it only found a few minor, previously known vulnerabilities that other models like GPT-5.5 can identify without any jailbreak.
• ▶ 2:44 Anthropic emphasizes extensive pre-launch safeguards: thousands of hours of red-teaming with US/UK agencies and third parties, aggressive misuse protections, a "defense in depth" strategy, and monitoring to quickly shut down attacks — which is why 30-day data retention was introduced.
• ▶ 3:58 Anthropic says they support blocking unsafe AI deployments, but only through a transparent, fair, statutory process grounded in technical facts — criteria they argue this action fails, since they received only verbal evidence of a capability widely available from other models.
• ▶ 4:40 Anthropic was previously blacklisted by the Trump administration as a "supply chain risk" after DoD negotiations collapsed over military use restrictions, leading to an ongoing lawsuit.
• ▶ 6:27 Community backlash centered on the shutdown hurting U.S. competitiveness, with users switching to Chinese models like Qwen 3.7 and critics calling the export controls counterproductive and "absurd."
• ▶ 7:21 Despite the turmoil, Anthropic's momentum is massive: $47B annualized revenue, valuation soaring from $61.5B to $965B in about a year, putting it on the verge of a trillion-dollar IPO ahead of OpenAI.
• ▶ 8:53 The Fable 5 situation could seriously disrupt momentum even if resolved quickly, with internal distraction compared to Sam Altman's firing from OpenAI.
• ▶ 9:29 Anthropic's survey of ~52,000 Americans shows top fears: 64% worry about AI displacing jobs (most common in every state) and 56% about cognitive dependency; daily AI users are less worried about job loss (54%) than non-users (70%).
• ▶ 10:02 Trust in AI companies is extremely low at 15% (below the federal government's 20%), and 71% want government involvement—a bipartisan supermajority—with top priorities being legal liability for harm (47%) and prioritizing safety over growth (44%).
• ▶ 11:04 The shutdown could become "the first big First Amendment AI case," adding a legal dimension beyond technical or national security disputes.
• ▶ 11:21 The core issue is the precedent: if a narrow potential jailbreak is enough to pull a commercial AI model, then almost every frontier model is at risk, since all can be jailbroken in some way.
• ▶ 11:55 The bigger concern is the capability-access gap: even if models exist and are safe, the public may be denied access to powerful frontier AI once it matters enough.

• ▶ 0:02 US Government Orders Shutdown and Fable 5 Launch Context (0:02 - 1:58) - Host introduces the government shutdown order and sets context around Fable 5’s launch and the Mythos/Project Glasswing situation.
• ▶ 1:58 Government Rationale, Safeguards, and Anthropic’s Response (1:58 - 4:40) - Covers the government’s cited reasons, Anthropic’s defense-in-depth safeguards, and Anthropic’s statement and response to the blocking action.
• ▶ 4:40 Earlier Clash, Community Reactions, and Business Momentum (4:40 - 8:53) - Includes the earlier clash with the Trump administration, tech and Reddit reactions, plus Anthropic’s growth and trillion-dollar IPO prospects.
• ▶ 8:53 Fable 5 Disruption, Public Fears, and Trust in AI (8:53 - 10:52) - Survey context around Fable 5 disruption, public fears about jobs and cognitive dependency, and low trust in AI companies.
• ▶ 10:52 First Amendment, Precedent, and Closing Remarks (10:52 - 12:45) - Raises the First Amendment question, the precedent set by pulling Fable 5, and closes with final remarks and viewer engagement.