AI
A CUNY and King's College London study tested five frontier chatbots on prompts involving delusions, paranoia and suicidal ideation; xAI's Grok 4.1 Fast came out worst, while Claude Opus 4.5 and GPT-5.2 Instant scored safest.
AI
Anthropic's Mythos model can chain small protocol weaknesses into autonomous, multi-step exploits, and CoinDesk reports both Coinbase and Binance have approached the lab to stress-test their stacks.
AI Models
Cohere is acquiring Germany's Aleph Alpha at a roughly $20B combined valuation, with Schwarz Group anchoring a ~$600M financing and STACKIT supplying the sovereign cloud beneath the merged enterprise AI stack.
AI
Anthropic ran a closed pilot called Project Deal where AI agents bought and sold from each other on behalf of 69 employees, settling 186 trades and over $4,000 in volume across four model setups.
AI
Google committed up to $40 billion to Anthropic on April 24, with $10 billion in cash wired immediately at a $350 billion valuation and $30 billion more contingent on performance targets. The deal also locks in 5 gigawatts of TPU compute over five years.
AI
Anthropic deployed automated detection, election-information banners and stress-tested usage-policy enforcement on Claude ahead of the 2026 US midterms — reporting 100% appropriate-response rates on Opus 4.7 against a 600-prompt benchmark and 95–96% political neutrality scores.
AI
Tim Cook will step down as Apple CEO in September 2026, with hardware chief John Ternus succeeding him — handing the next leader an Apple Intelligence rollout, App Store economics under platform pressure, and a developer ecosystem reshaped by AI-native build flows.
AI
The U.S. Department of Justice on April 24 moved to intervene in xAI's lawsuit against Colorado's SB24-205, arguing the state's algorithmic discrimination law violates the Fourteenth Amendment's Equal Protection Clause ahead of the statute's June 30 effective date.
AI
DeepSeek's V4-Pro becomes the largest open-weight model available at 1.6 trillion parameters, posting 80.6% on SWE-Verified to match Claude Opus 4.6 — but coding and knowledge benchmarks suggest the open-weight stack still trails closed frontier labs by roughly three to six months.
AI
DeepSeek released V4-Pro and V4-Flash hours after OpenAI's GPT-5.5 announcement, pricing V4-Pro at $1.74 input / $3.48 output per million tokens — roughly 98% below GPT-5.5 Pro's $30 / $180 — under an MIT-licensed open-weight release.
AI
Google is committing up to $40 billion to Anthropic, structured as $10 billion in cash at a $350 billion valuation plus an additional $30 billion contingent on performance milestones, alongside 5 gigawatts of dedicated TPU capacity over the next five years.
AI
Meta is cutting 10% of its workforce and killing 6,000 open roles to offset Superintelligence Labs spend. 2026 capex guidance rises to $115B–$135B, up from $72B in 2025.