📖 2 min read
Anthropic Drops Claude Opus 4.7 — Claims 92% Honesty Rate
Anthropic released Claude Opus 4.7 on Thursday, a hybrid reasoning model that the company says is significantly less likely to hallucinate or engage in sycophancy than any prior Claude model — or competing frontier models.
According to the system card, Opus 4.7 shows “large reductions in the rate of important omissions, and moderate improvements in factuality.” The 92% honesty benchmark is a notable jump, and Anthropic is clearly positioning safety and reliability as its competitive edge heading into the second half of 2026.
📧 Want more like this? Get our free The Ultimate AI Tool Database: 200+ Tools Rated & Ranked — Downloaded 5,000+ times
White House Races to Get Anthropic Mythos Access Amid Security Fears
The U.S. government is in active talks with Anthropic to deploy its frontier model, Claude Mythos Preview, across major federal agencies. According to Bloomberg and Reuters, the White House is pushing for access despite ongoing concerns that Mythos could uncover – and potentially create – cybersecurity vulnerabilities at unprecedented speed.
Anthropic CEO Dario Amodei met with White House Chief of Staff Susie Wiles last week to discuss the implications. The irony is hard to miss: the government wants the model precisely because it is powerful enough to be dangerous.
DeepSeek Raising First External Round at $10B Valuation
Chinese AI startup DeepSeek is in talks to raise at least $300 million at a $10 billion valuation, according to The Information. This marks the company’s first external fundraise – a significant milestone for the lab that shook up the industry earlier this year with cost-efficient open models.
📧 Want more like this? Get our free The Ultimate AI Tool Database: 200+ Tools Rated & Ranked — Downloaded 5,000+ times
The round signals strong investor confidence that the cost-efficient approach to frontier AI has serious legs. DeepSeek remains one of the most closely watched AI companies outside the U.S.
Google Teams Up with Marvell on Next-Gen AI Inference Chips
Alphabet’s Google is in talks with Marvell Technology to co-develop two new chips purpose-built for running AI models more efficiently. One is a memory processing unit designed to pair with Google’s TPUs, and the other is a new TPU built specifically for inference workloads.
Marvell shares jumped on the news Monday. The move signals Google is doubling down on custom silicon rather than relying solely on Nvidia – a trend that could reshape the AI chip landscape as inference costs become the next major battleground.
📧 Want more like this? Get our free The Ultimate AI Tool Database: 200+ Tools Rated & Ranked — Downloaded 5,000+ times
UK Launches £500M Sovereign AI Fund
The UK government officially launched its £500 million ($675M) Sovereign AI Fund last week, aimed at scaling homegrown AI startups across model development, agentic AI, and drug discovery. Tech Secretary Liz Kendall announced the fund at autonomous driving company Wayve’s offices.
UK AI startups raised £6 billion in venture capital last year, and have already pulled in more than half that figure in just the first three months of 2026. The fund will operate like a state-backed VC, making direct investments in British AI firms.
Stay sharp. Follow BetOnAI.net for AI investment analysis and AiToolCrush.com for hands-on AI tool reviews.