Google DeepMind abandoned the Gemini 2.5 Pro base architecture and pushed the 3.5 Pro launch to July 17, opting for a complete rebuild to close performance gaps in mathematical reasoning and SVG generation before competing releases from OpenAI and Anthropic.
Tag: artificial intelligence
Posted inTech
EU AI Act’s GPAI penalty powers take effect August 2 — what changes in 33 days
On August 2 the EU AI Act activates penalty enforcement powers over AI model providers — up to €15m or 3% of global turnover — alongside new chatbot transparency rules and revised high-risk system deadlines.
Posted inTech
Anthropic says Alibaba used 25,000 fake accounts to run 28.8 million Mythos queries for Qwen training
Anthropic disclosed that operators linked to Alibaba’s Qwen lab ran 28.8 million interactions through 25,000 fraudulent accounts over 45 days, harvesting its Mythos model’s outputs to train rival AI systems.
Posted inTech
Reflection AI takes $6.3bn SpaceX compute deal as AI infrastructure concentrates
Reflection AI will pay SpaceX $150 million a month for Nvidia GPU access at Colossus 2, in a contract worth up to $6.3 billion through 2029.
Posted inMarkets
Micron guides Q4 revenue to $50 billion as AI memory demand outruns supply
Micron reported Q3 FY2026 revenue of $41.46 billion — smashing guidance by $8 billion — and guided Q4 to $50 billion as AI memory demand outpaces chip supply.
Posted inTech
Forty-two state attorneys general subpoena OpenAI days after its IPO filing
A coalition of 42 states subpoenaed OpenAI just four days after its IPO filing, targeting ChatGPT’s health data practices, treatment of minors, and a design flaw called model sycophancy.
Posted inTech
EU picks Domyn’s EUROPA to build a 400-billion-parameter open-source AI in all 24 EU languages
The EU selected Italian startup Domyn’s EUROPA consortium to build a 400-billion-parameter open-source AI model covering all 24 official EU languages.
Posted inTech
Anthropic and DeepMind CEOs call for US-led global AI coalition at G7
Anthropic’s Dario Amodei and Google DeepMind’s Demis Hassabis called for US leadership on global AI standards at the G7 summit in Évian on June 17.
Posted inTech
London Tech Week closes with £6bn in AI pledges and a national compute push
London Tech Week 2026 concluded with £6 billion in AI pledges, including AMD’s £2bn five-year commitment and a UK government £1.1bn AI Hardware Plan anchored by a £750m national supercomputer.
Posted inTech
German court declares Google liable for false AI Overview claims
Munich’s Regional Court found Google directly liable for false AI Overview statements, ruling the summaries are Google’s own speech and not protected search-result listings.