AI news · Saturday, June 27, 2026

Anthropic regains access to its flagship cybersecurity model Mythos 5

The Trump administration has partially lifted the ban on Anthropic’s most advanced cybersecurity model, Mythos 5, following two weeks of high-stakes negotiations. While the government remains cautious—the consumer-facing Fable 5 model is still effectively off-limits—it has authorized access for a specific group of over 100 trusted U.S. government agencies and critical infrastructure providers. This move comes after the White House initially blocked all non-U.S. nationals from accessing the models, a decision that disrupted the work of Anthropic’s own staff. OpenAI has faced similar scrutiny, noting that such government-mandated access processes should not become the industry default. Meanwhile, Satya Nadella is pushing a different philosophy, arguing that every company should build its own custom AI models rather than relying on a small handful of centralized frontier systems, which he warns creates an economic single point of failure.

Corporate stability is also showing cracks as the AI boom reshapes the tech landscape. Domo, once a $2.8 billion data-analytics darling, is currently fighting for its survival, missing loan covenants and facing a massive 80% stock decline over the past year. The company’s struggles—compounded by executive departures, a CEO’s DUI arrest, and stiff competition from AI agents that perform the clerical work its platform was designed to automate—have left the firm seeking a buyer before the end of July. This broader market anxiety has led to what analysts call a 'mega rotation,' where capital is fleeing high-valuation tech stocks for value and cyclical sectors. Investors are increasingly wary of an AI bubble, with the S&P 500 near historic highs, leading some firms to suggest that clients diversify away from the U.S. tech-heavy indices.

These market shifts are creating real-world friction for workers. Google employees report leaving for startups not just for equity, but because the AI-driven restructuring is making Big Tech feel less like a safe haven. Meanwhile, the pressure to maintain in-office presence is intensifying; SaaStr founder Jason Lemkin has declared he will only invest in companies that demand six-day work weeks, labeling remote work a non-starter. Even the hiring process has become a high-stakes performance, with candidates at startups like Perplexity and Kalshi resorting to grand gestures—like sending engraved clocks or flying across the country at their own expense—just to get an interview.

The quick hits

Anthropic is restoring access to its Mythos 5 model for select government and infrastructure partners after a two-week regulatory freeze — this marks a thaw in the government's direct intervention in AI model rollouts.
Data-analytics firm Domo is actively seeking a buyer to avoid collapse — a reminder that traditional software companies are struggling to pivot as AI agents begin automating their core services.
Investors are shifting capital away from big tech into value sectors to hedge against a potential AI bubble — the market is showing fatigue as valuations hit levels not seen since the dot-com era.
Startup founders are increasingly demanding six-day in-office work weeks for their teams — the era of the relaxed remote tech job is being aggressively challenged by high-pressure investor mandates.

Anthropic regains access to its flagship cybersecurity model Mythos 5

The quick hits

Sources