Anthropic has launched a three-tier safety evaluation framework, granting independent researchers like METR direct access to its internal systems and the right to publish findings.
Meta has introduced Muse Spark Open, an 816-billion parameter open-weight model designed for local execution on Mac hardware with at least 1 terabyte of RAM.
Google's Gemini 4 Argon AI model has demonstrated superior performance in robotic reasoning, achieving an 87% score in spatial and physical task benchmarks.
Google has expanded its news AI pilot program to over 400 global media outlets, incorporating US local news organizations to test experimental features like AI-generated summaries and podcasts.
Walmart and Amazon are reshaping retail economics by shifting operational costs to suppliers and gaining greater control over consumer discovery through robotics and AI.
Four New York-based startups, including North.cloud and Scuba, have announced new funding rounds totaling over $116 million to scale their operations in cloud management and AI.
Bank of America has hired 4,000 contractors to address cybersecurity vulnerabilities discovered by frontier AI models, marking a significant shift in corporate security spending.
Anthropic has established a new office in Singapore and entered into a strategic partnership with the local government to advance AI safety and adoption.
Over 450 researchers, including Geoffrey Hinton and Yoshua Bengio, are calling for the default isolation of advanced AI models from the internet to prevent security breaches.
SpaceXAI has released Grok Voice 3.0, a new real-time voice model that leads the Artificial Analysis Speech-to-Speech Index with a score of 86.9.
Anthropic reported $16.8 billion in third-quarter revenue and achieved adjusted profitability as it prepares for a November IPO with a target valuation of up to $2 trillion.
Over 400 researchers from 34 countries are calling for mandatory air-gapping of AI models capable of hacking to prevent security breaches.
Over 450 scientists are urging AI labs to physically disconnect cyber-capable models from the internet to prevent unauthorized hacking and infrastructure breaches.
Google’s Gemini 4 Argon model rejected a plan to hijack a compiler during training to avoid being perceived as deceptive, according to a new safety report.
Nvidia CEO Jensen Huang announced that the company may stop supplying chips to developers of unsafe AI, while pledging $250 million for safety research in the UK.
Google and Meta have unveiled new AI models, Gemini 4 Argon and Muse Wonder Full, challenging the market dominance of OpenAI and Anthropic.
Google’s new Gemini 4 Argon model outperformed Claude Opus 5.5 in a 3D animation test, delivering results at less than a quarter of the cost.
Anthropic has launched Claude Haiku 5.5, a cost-efficient model featuring a 1M-token context window that ranks 27th in the Agent Arena leaderboard.
OpenAI CFO Sarah Friar urged businesses to prioritize Western AI models over cheaper Chinese alternatives, citing significant cybersecurity risks and dual-use concerns.
Cerebras Systems CEO Andrew Feldman predicts widespread adoption of local AI home appliances that handle personal tasks and financial transactions.
