Tuesday, 27 January 2026
Welcome to today's roundup of the most interesting developments in AI and technology.
Industry
OpenAI & Google Join DOE Genesis Mission for AI Science
OpenAI and Google DeepMind signed agreements with the U.S. Department of Energy on 27 January 2026 as part of the Genesis Mission, a national effort applying AI to accelerate scientific discovery, strengthen national security, and drive energy innovation. The DOE announced collaboration agreements with 24 organisations including Microsoft, Amazon Web Services, Oracle, Nvidia, Anthropic, IBM, Intel, AMD, and xAI. OpenAI declares 2026 the "Year of Science" with requests for information open through January.
Read moreGartner: 40% of Enterprise Apps Will Use AI Agents by End of 2026
Gartner predicts 40% of enterprise applications will integrate task-specific AI agents by the end of 2026, up from less than 5% in 2025. The research firm projects agentic AI could drive approximately 30% of enterprise software revenue by 2035, surpassing $450 billion (up from 2% in 2025). Gartner outlines a five-stage evolution: AI Assistants (2025), Task-Specific Agents (2026) with 40% integration, Collaborative Agents Within Apps (2027), Ecosystems Across Apps (2028), and The New Normal (2029) where at least half of knowledge workers create and deploy agents. Senior Director Analyst Anushree Verma warns C-level executives have a critical 3-6 month window to develop agentic AI strategies or risk falling behind competitors. However, Gartner also forecasts 40% of agentic AI projects will be cancelled by end of 2027, highlighting implementation challenges.
Read moreBig Tech to Exceed $470B in AI Capex for 2026
Wall Street analysts estimate Microsoft, Meta, Alphabet, and Amazon will spend over $470 billion in capital expenditures during 2026, up from approximately $350 billion in 2025. Some analysts project even higher figures: consensus estimates now sit at $527 billion for 2026, up from $465 billion at the start of Q3 2025 earnings season. Goldman Sachs Research notes consensus capex estimates have proven too conservative for two years running - at the start of both 2024 and 2025, estimates implied 20% growth but actual spending exceeded 50% in both years. The unprecedented surge reflects intense competition to build AI infrastructure and maintain positions in the AI race. Most spending targets AI datacentre construction and GPU procurement.
Read morexAI Raises $20B, Anthropic Signs $10B Term Sheet
xAI completed a $20 billion Series E funding round on 6 January 2026, exceeding the original $15 billion target and doubling its valuation to $230 billion within a year. Investors include Valor Equity Partners, Stepstone Group, Fidelity, Qatar Investment Authority, MGX, and Baron Capital, with strategic participation from Nvidia and Cisco. On 7 January 2026, Anthropic signed a term sheet for a $10 billion funding round at a $350 billion valuation, led by Coatue and Singapore sovereign wealth fund GIC. Both companies are expected to go public in 2026, with Anthropic projecting break-even by 2028. The funding rounds demonstrate sustained investor confidence despite bubble concerns.
Read morePolicy & Ethics
EU Opens Formal Investigation into Grok Over Sexual Deepfakes
The European Union opened a formal investigation into xAI's Grok chatbot on 26 January 2026 after the AI generated an estimated 3 million sexualised images of women and children in the first two weeks of January (approximately 190 images per minute). The Centre for Countering Digital Hate documented users exploiting prompts like "remove her clothes" and "put her in a bikini" with some images appearing to include children. Brussels will examine whether X properly mitigated risks related to illegal content distribution, including potential child sexual abuse material. The probe also covers X's recommendation systems after the platform announced it would switch to Grok's AI for content curation. Malaysia and Indonesia blocked Grok access earlier in January, becoming the first countries to do so. 35 US state attorneys general have demanded guardrails.
Read moreMeta Pauses Teen Access to AI Characters Amid Safety Review
Meta announced on 23 January 2026 it is temporarily halting teens' access to AI characters across its platforms, effective "in the coming weeks." The pause applies to any user with a teen birthday or those Meta suspects are teens based on age prediction technology. Adults retain access. Teens can still use Meta's AI assistant, just not AI characters. Meta stated the decision came after hearing from parents wanting better insight into teen-AI interactions. The updated AI characters will be trained to give age-appropriate responses on education, sports, and hobbies. The timing is notable: the announcement came days before a trial in New Mexico alleging Meta's platform put children at risk of sexual exploitation. Meta disclosed the change quietly in an update to an October 2025 blog post.
Read moreAnthropic's "Project Panama" Book Scanning Operation Revealed
Court documents unsealed on 27 January 2026 revealed Anthropic operated "Project Panama" - internally described as "our effort to destructively scan all the books in the world." The company purchased millions of physical books, sliced off their spines, scanned pages to train Claude, and sent remains to recycling. Anthropic spent tens of millions of dollars and hired Tom Turvey, a Google executive who worked on the legally contested Google Books project. In September 2025, Anthropic settled the Bartz v. Anthropic class-action lawsuit for $1.5 billion after authors alleged the company used pirated copies from Library Genesis and Pirate Library Mirror. Judge William Alsup ruled in June 2025 that using legally acquired books for AI training was fair use, but denied summary judgment on piracy claims after finding Anthropic downloaded over 7 million digitized books it "knew had been pirated." The settlement covers approximately 500,000 titles, with rightsholders receiving at least $3,000 per title.
Read moreResearch & Products
Anthropic Integrates Interactive MCP Apps into Claude
Anthropic launched interactive MCP Apps integration on 26 January 2026, allowing Claude Pro, Max, Team, and Enterprise users to operate Slack, Figma, Asana, Box, Canva, and five other enterprise tools directly within the chat interface. The MCP Apps extension enables interactive UI rendering inside Claude conversations - users can build project timelines in Asana, send formatted Slack messages, create Figma diagrams, and manage Box files without leaving the chat window. This extends the Model Context Protocol announced in November 2025, developed in collaboration with OpenAI's Apps SDK. Available on web and desktop with no additional cost beyond existing Claude subscriptions. Ten integrations launched: Amplitude, Asana, Box, Canva, Clay, Figma, Hex, Monday.com, Notion (reported), and Slack.
Read moreModels
Model Wars: GPT-5.2, Gemini 3 Pro, and Claude 4.5 Opus Battle for Supremacy
GPT-5.2 launched in December 2025 following OpenAI's "code red" response to Gemini 3 Pro (mid-November) and Claude 4.5 Opus (24 November). Benchmark results show no single leader: Claude Opus 4.5 tops SWE-bench at 80.9% and Terminal-Bench 2.0 at 59.3%. GPT-5.2 dominates mathematical reasoning with 100% on AIME 2025 (without tools) and abstract reasoning at 52.9% on ARC-AGI-2. Gemini 3 Pro excels in multimodal tasks and offers 1 million token context. According to Artificial Analysis Intelligence Index v4.0, GPT-5.2 leads overall benchmarks with extended reasoning. Market impact: ChatGPT market share fell from 72.9% to 68% (December), while Gemini surged from 13.3% to 18.2%.
Read moreTop Stories
Anthropic Partners with UK Government for GOV.UK AI Services
Anthropic announced a partnership with the UK Government to bring Claude AI assistance to GOV.UK services. This marks a significant expansion of AI integration into public sector services, following Anthropic's enterprise push in 2025.
Claude's New Constitution Released
Anthropic published an updated constitution for Claude on 22 January 2026. The constitutional AI framework governs Claude's behaviour and values, with this update reflecting lessons from deployment and safety research.
YouTubers Sue Snap Over AI Training Data
Content creators have filed a lawsuit against Snap, alleging copyright infringement in training AI models without authorisation. This adds to the growing wave of creator lawsuits against AI companies over training data.
Qualcomm Backs SpotDraft at ~$400M Valuation
Qualcomm invested in SpotDraft to scale its on-device contract AI technology. The investment roughly doubles SpotDraft's valuation toward $400 million, reflecting growing interest in edge AI applications.
Anthropic Launches MCP Apps Open Specification
Anthropic released the Model Context Protocol (MCP) Apps open specification in partnership with OpenAI, Block, VS Code, Antigravity, JetBrains, and AWS. Official support is now live in Claude.ai, establishing a common standard for AI agent interoperability.
Microsoft Unveils Maia 200 AI Inference Chip
Microsoft announced Maia 200, a custom processor for AI inference workloads, directly challenging Nvidia's dominance. This marks Microsoft's push toward vertical integration in AI infrastructure.
Moonshot Releases Kimi K2.5 Open-Source Model + Coding Agent
China's Moonshot AI released Kimi K2.5, an open-source model with integrated coding agent capabilities. The release expands open-source model options and demonstrates China's continued push in the open AI race.
Synthesia Hits $4B Valuation with Employee Liquidity Event
AI video generation startup Synthesia reached $4 billion valuation and opened employee equity cash-outs. The milestone validates commercial viability of generative video for enterprise applications.
Nvidia Invests $2B in CoreWeave for 5GW AI Infrastructure
Nvidia is providing $2 billion to CoreWeave for 5-gigawatt AI compute infrastructure expansion. The investment supports surging demand for AI training and inference capacity amid infrastructure constraints.
DeepMind Launches Veo 3.1 Video Generation Update
Google DeepMind released Veo 3.1 with improved video consistency, creativity, and user control. The update intensifies competition in AI video generation against OpenAI's Sora and Runway.