Anthropic AI Updates: April 14, 2026

1. Jack Clark at Semafor: World Must “Get Ready” for Mythos-Class Capabilities

Anthropic. Co-founder Jack Clark, speaking at the Semafor World Economy event in Washington on April 13, disclosed that Anthropic is actively engaging with the Trump administration about Mythos and future models. Clark warned that equally capable models will emerge from competitors within months and from Chinese open-source projects within 18 months, declaring the world must “get ready for more powerful systems.” He noted that Anthropic’s Pentagon contract dispute should not overshadow its commitment to national security. Source

2. OpenAI Memo Accuses Anthropic of Inflating Revenue by $8 Billion

Anthropic. An internal OpenAI memo from CRO Denise Dresser, leaked over the weekend and widely reported on April 13, accuses Anthropic of overstating its $30B run rate by roughly $8 billion through gross revenue accounting of cloud-partner sales. The memo criticizes Anthropic’s strategy as built “on fear, restriction, and the idea that a small group of elites should control AI,” while acknowledging Anthropic’s significant lead among enterprise customers in coding. Source

3. Cybersecurity Industry Debates Mythos: “The Real Problem Is Fixing, Not Finding”

Anthropic. Contrast Security CISO David Lindner challenged the Mythos panic narrative, arguing that vulnerability discovery was never the bottleneck since over 99% of vulnerabilities Mythos found remain unpatched. Security expert Zach Lewis countered that the model could democratize exploitation for threat actors lacking coding expertise. The debate frames a broader industry question about whether Mythos is a net positive or negative for cybersecurity. Source

4. Claude Code v2.1.105 Adds Worktree Path Switching and PreCompact Hooks

Anthropic. Claude Code v2.1.105 shipped with a path parameter for the EnterWorktree tool to switch into existing worktrees, PreCompact hook support that can block compaction, and background monitor support for plugins via a monitors manifest key. The release also improves stalled API stream handling with a 5-minute abort and non-streaming retry, upgrades /doctor with status icons and auto-fix, and includes over 20 bug fixes. Source

5. TechCrunch: At the HumanX Conference, Everyone Was Talking About Claude

Anthropic. TechCrunch reported that at the HumanX AI conference, Claude dominated conversations with attendees and vendors consistently naming it as the most popular chatbot. One vendor stated his team preferred Claude, claiming OpenAI had “fell off.” The article attributed declining OpenAI perception to product cancellations, advertising in ChatGPT, and the contentious New Yorker profile of Sam Altman, while noting Anthropic continues gaining enterprise ground. Source

6. Michael Burry Calls Anthropic the “Bigger AI Winner,” Continues Shorting Palantir

Anthropic. Investor Michael Burry maintained put options against Palantir while declaring Anthropic the bigger AI winner, noting “Anthropic went from $9B to $30B in months, it took Palantir 20 years to get to $5.” He argued Anthropic offers “the easier, cheaper, intuitive solution for businesses” and cited data showing 73.3% of new businesses choose Anthropic over OpenAI. Palantir stock fell over 13% in five days following Burry’s post. Source