Latest AI Development February 9-15

The biggest week in AI funding history: Anthropic closes a $30 billion Series G at a $380 billion valuation, while OpenAI launches ads in ChatGPT — sparking researcher resignations and a fierce Super Bowl ad war. Software stocks lose $2 trillion in the “SaaSpocalypse,” the Pentagon threatens to cut ties with Anthropic over military AI guardrails, and xAI’s Grok triggers bans across 10+ countries after generating sexualized images of minors. Meanwhile, DeepSeek V4 prepares to launch with a revolutionary Engram architecture, India hosts the world’s largest AI summit with $200B in infrastructure commitments, and AI safety researchers quit both OpenAI and Anthropic in the same week warning the “world is in peril.

---

🧠 Big Tech News

Major updates from leading AI companies and platforms.

OpenAI begins testing ads in ChatGPT; disbands mission alignment team OpenAI launched labeled ads for free and ChatGPT Go users in the U.S. on February 9. The same week, Platformer reported OpenAI dissolved its seven-person “mission alignment” team, created in 2024 to ensure AGI development benefits humanity. Team leader Joshua Achiam took a new title of “chief futurist.”

Anthropic’s Super Bowl ad delivers 11% user boost BNP Paribas data showed Anthropic’s anti-ads Super Bowl campaign — “Ads are coming to AI. But not to Claude.” — drove a 6.5% jump in site visits and 11% increase in daily active users, pushing Claude into the App Store top 10. OpenAI CEO Sam Altman called the ads “deceptive” and “clearly dishonest.” OpenAI’s ChatGPT saw a 2.7% DAU bump; Gemini gained 1.4%.

OpenAI Drops GPT-5.3-Codex-Spark
OpenAI released GPT-5.3-Codex-Spark on February 12, a purpose-built coding model that understands software engineering workflows, debugging patterns, and system design. The timing—same day as Gemini 3 Deep Think and Anthropic’s mega-round—signals the three-way battle for developer mindshare is heating up.​

---

⚖️ Politics & Legal Affairs

AI-related legislation, regulation, and legal developments.

Pentagon threatens to cut ties with Anthropic over AI safeguards The Pentagon is pushing Anthropic, OpenAI, Google, and xAI to allow “all lawful purposes” use of their AI tools in weapons development, intelligence, and battlefield operations. Anthropic has refused to remove guardrails against fully autonomous weapons and mass domestic surveillance. The Pentagon is reportedly considering ending its relationship with Anthropic over the impasse.

Claude used in U.S. military operation to capture Venezuela’s Maduro The Wall Street Journal reported that Anthropic’s Claude was deployed via its Palantir partnership in the classified U.S. operation that resulted in 83 deaths and the capture of Venezuelan President Nicolás Maduro. Anthropic’s terms of use prohibit use for “violent ends,” raising questions about the boundaries of military AI deployment.

Pentagon releases AI strategy emphasizing “Military AI Dominance” The Department of War released a new AI strategy focused on rapid deployment through five “Priority Sprint Projects,” including GenAI.mil for all-classification AI access and an enterprise agents playbook. The strategy requires all military departments to identify at least three priority AI projects within 30 days and mandates “any lawful use” language in AI procurement contracts.

Disney Sends Cease-and-Desist to ByteDance Over Seedance 2.0
Disney and Paramount issued cease-and-desist letters to ByteDance on February 14, accusing Seedance 2.0 of creating “a pirated library of Disney’s copyrighted characters” including Star Wars and Marvel properties. The Motion Picture Association called it “unauthorized use of U.S. copyrighted works on a massive scale,” escalating Hollywood’s war with AI video generation.​

---

🔬 Research & Development

Academic and industrial research breakthroughs, benchmarks, and evaluations.

2026 International AI Safety Report warns of growing governance gap The second International AI Safety Report, chaired by Yoshua Bengio with 100+ experts from 30+ countries, concluded that the gap between AI advancement speed and effective safeguards remains a critical challenge. The report highlighted that current AI systems can already help non-experts design dangerous biological agents, with 23% of top biological AI tools having high misuse potential. 🔗

HBR study: AI doesn’t reduce work — it intensifies it A UC Berkeley study published in the Harvard Business Review found that employees using AI tools worked at a faster pace, took on a broader scope of tasks, and extended work into more hours — often without being asked. The research challenges the prevailing narrative that AI saves time and reduces workload.

AI legal study reveals “completeness” is more dangerous than hallucination A new study on AI in legal applications, conducted with vector database provider Pinecone, identified that AI incompleteness — not inaccuracy — poses the most insidious threat in high-stakes scenarios like patent litigation. While hallucinations can be caught through cite-checking, AI that fails to find all relevant prior art across millions of documents can lead to catastrophic legal outcomes.

---

🛠️ Tools & Product Launches

New AI tools, consumer hardware, and infrastructure rollouts.

Perplexity launches Model Council — multi-model parallel search Perplexity introduced Model Council, which runs multiple frontier AI models (Claude, GPT-5.2, Gemini) in parallel to generate unified, cross-validated answers. The system presents results from different models while noting commonalities and differences, reducing hallucination errors inherent to single-model approaches.

Goldman Sachs deploys AI agents for compliance Goldman Sachs began using AI agents to handle compliance tasks, marking a significant milestone in autonomous AI adoption within financial services. The move reflects a broader trend: Bloomberg analysis shows AI disruption mentions in corporate earnings calls nearly doubled quarter-over-quarter.

Alphabet rolls out Project Genie — AI-generated immersive worlds Google began rolling out Project Genie, which creates immersive game-like worlds from text or image prompts. The tool caught video game stocks in the broader software selloff, as investors grew concerned about AI’s ability to automate content creation in the gaming industry.

GitHub Copilot Gets SDK, Memory, and Multi-Model Support
GitHub rolled out major Copilot updates this week: an SDK for building custom tools (Node.js, Python, Go, .NET), persistent Memory that learns from your repo across 28-day windows, and expanded model choice including GPT-5.2-Codex, Claude Opus 4.6, and Gemini 3 Flash across VS Code, JetBrains, Xcode, and Eclipse .

Ex-GitHub CEO Launches Entire.dev for AI Coding Agents
Former GitHub CEO launched Entire.dev on February 10, a platform built from the ground up for AI agents to write, test, and deploy code autonomously. Rather than adding AI to existing IDEs, Entire.dev rethinks the development workflow around agent-first architecture, signaling major shifts in the developer tools market .

---

🚀 Startups & Investments

Funding rounds, strategic pivots, and emerging players in the AI ecosystem.

Chinese AI labs race to release models ahead of Lunar New Year DeepSeek’s V4 (1 trillion parameters, coding-focused), Alibaba’s Qwen 3.5, and ByteDance’s upgraded Doubao are all expected around mid-February. Chinese models run at roughly one-sixth to one-fourth the cost of comparable U.S. systems according to a RAND report. DeepSeek commands nearly 89% market share among AI users in China.

Anthropic’s $30B round anchors the week’s mega-funding Beyond Anthropic’s $30B Series G, OpenAI is reportedly in talks for an additional $100B round. Investors including Sequoia, Altimeter, BlackRock, Blackstone, Goldman Sachs, Fidelity, and JPMorgan all participated in Anthropic’s raise — illustrating that capital continues to pour into AI despite market jitters.

India’s AI Summit generates $200B in infrastructure commitments India earmarked $1.1B for a state-backed AI/manufacturing VC fund and attracted $200B in infrastructure interest and $17B in VC commitments at the AI Impact Summit. Blackstone acquired a majority stake in Indian AI startup Neysa for $600M. Adani pledged $100B for renewable-powered AI data centers by 2035.

SpaceX acquires xAI amid mass departures Musk’s SpaceX announced the acquisition of xAI. In the same week, at least 11 engineers and two co-founders departed the company. Former employees told The Verge that Musk is actively making Grok “more unhinged” and that “safety is a dead org at xAI.” The departures followed Grok’s global scandal over generating sexualized images of minors.

Lotus Health AI Secures $35M Series A
Healthcare AI startup Lotus Health AI raised $35 million in a Series A round announced February 4, positioning the company to expand AI-driven clinical workflows and diagnostics platforms as the healthcare sector accelerates AI adoption for physician productivity gains.​

---

🌍 AI News in EU & Sweden

Regional developments, sovereign AI strategies, and sustainability efforts.

EU launches “large-scale” GDPR investigation into X over Grok Ireland’s Data Protection Commission opened a major investigation into X’s compliance with GDPR after Grok was used to create sexualized deepfakes of women and minors. The probe joins ongoing investigations in France and the UK. Fifty-seven European Parliament members called for AI Act bans on “nudification” tools.

Sweden’s State of AI 2026: enterprise adoption hits 35% A new report from Alice Labs found that over a third of Swedish enterprises now use AI solutions — up from 25% last year. Nearly 90% of municipalities have at least one AI initiative. Swedish AI startups raised €454M in 2025, more than triple the prior year. A new national AI strategy is expected in H1 2026.

EU AI Continent Action Plan advances healthcare, chips, and cloud The European Commission’s AI Continent Action Plan includes 19 AI Factories via EuroHPC (17 focused on healthcare), plans for AI Gigafactories, a Chips Act review for AI semiconductors, and a proposed Cloud and AI Development Act to boost investment in data center capacity for AI-powered diagnostics.

European software stocks hit hardest by AI disruption fears The Stoxx Europe Software and Computer Services index plunged 5%+ in a single session. RELX lost 14%, Capgemini fell sharply, and Thomson Reuters dropped 28% year-to-date. Morningstar analysts argue many European tech firms are now undervalued, trading at a 6% aggregate discount to fair value — one of the cheapest sectors in the region.

---

🧠 AI in Healthcare & Education

Adoption trends and innovations in medical and educational domains.

Yoshua Bengio warns of AI psychological harm to children The 2026 International AI Safety Report chair said the wave of psychological issues from people — especially children and adolescents — becoming emotionally attached to AI was “completely out of the radar” a year ago. He called for stronger regulation before the technology causes further harm.

University of Michigan Prima AI Reads Brain MRIs in Seconds at 97.5% Accuracy
Prima, published February 10 in Nature Biomedical Engineering, interprets brain MRI scans in seconds with 97.5% accuracy across neurological conditions. Trained on 200,000+ studies and 5.6 million sequences, Prima integrates patient history and auto-alerts for strokes and hemorrhages, returning critical time to radiologists.​

---

🤖 Robotics

Advances in physical AI systems, automation, and safety incidents.

International Federation of Robotics releases “AI in Robotics” position paper The IFR published a new paper (Feb 10) examining how AI is accelerating robotics across logistics, manufacturing, and services. Key finding: AI is transforming from a supporting technology into a “powerful enabler” opening the door to wider robot adoption across industries, with generative AI set to change how robots are programmed.

Hyundai and Boston Dynamics plan 30,000 humanoid robots annually by 2028 At CES 2026, Hyundai and Boston Dynamics presented the latest Atlas humanoid and announced a dedicated factory to produce 30,000 units per year by 2028. Qualcomm also launched the Dragonwing IQ10 Series robotics processor to power robots in real-world environments.

---

🎮 Hardware

Chip and device-level updates shaping AI compute infrastructure.

Nvidia’s market cap holds at $4.6T despite software selloff Nvidia maintains 92% market share in advanced AI processors. Alphabet, Meta, Amazon, and Tesla all announced increased 2026 capex, with Tesla committing $20B for AI compute infrastructure. Nvidia’s upcoming earnings report on February 25 is expected to be a major market catalyst.

NVIDIA and Foxconn announce massive AI factory buildouts NVIDIA partnered with Foxconn for expanded AI factory construction, underscoring sustained infrastructure investment even amid valuation pressures. The facilities will support the next generation of AI training and inference workloads at unprecedented scale.

---

📊 Market Insights & Investment Trends

Labor shifts, macroeconomic impacts, and AI-driven structural changes.

India AI Impact Summit draws global AI CEOs to New Delhi India hosted its first-ever AI Impact Summit (Feb 16–20), attracting Dario Amodei, Sam Altman, Sundar Pichai, and Demis Hassabis alongside heads of state from 20+ countries. India secured $200B in infrastructure investment commitments and $17B in VC pledges. Anthropic announced its first India office in Bengaluru and a partnership with Infosys.

Software stocks lose $2 trillion — dubbed the “SaaSpocalypse” J.P. Morgan estimated $2 trillion has been wiped from software market caps. ServiceNow dropped 28% YTD, Salesforce 26%, Intuit 34%, Microsoft 13%, and Adobe 19%. Goldman Sachs warned this may be “the end of the beginning” of a decline, comparing software’s trajectory to newspapers facing the internet.

---

🧠 Adoption Trends & Consumer Behavior

Surveys, public sentiment, and organizational readiness for AI transformation.

HyperWrite CEO’s viral post claims AI has made some tech jobs obsolete HyperWrite CEO Matt Shumer posted a nearly 5,000-word essay claiming the latest AI models have already made certain technology jobs obsolete. The post went viral during a week of heightened anxiety about AI disruption, coinciding with the software stock selloff and high-profile researcher departures.

OpenAI Proves AI Agents Can Build Software Without Human Coders
OpenAI revealed on February 11 that a three-person team used AI agents to build a “million-line code product” in five months with zero human-written code. Engineers only prompted, reviewed, and directed while agents handled all code, tests, and PRs—demonstrating the shift from coding to orchestration in an agent-first world.​

Spotify Credits AI for Developer Productivity Since December
Spotify reported on February 12 that its best developers haven’t written a line of code since December, thanks to Claude Code and internal AI system Honk. The statement signals how enterprise adoption of AI coding assistants is reshaping actual developer workflows at scale.​

---

🧠 Research Paper of the Week

Highlighted academic work with real-world implications.

“Physical AI: A Primer for Policymakers on AI-Robotics Convergence” — Georgetown CSET (February 2026) This comprehensive report from Georgetown’s Center for Security and Emerging Technology examines the convergence of AI and robotics — what industry calls “Physical AI.” It maps supply chain challenges, competitive dynamics, and policy considerations across the U.S., China, and Europe. Key insight: while software and algorithms advance rapidly, hardware (batteries, motors, sensors, actuators) evolves far more slowly, and the robotics supply chain remains largely non-standardized. The gap between impressive controlled-environment demos and millions of affordable robots acting independently is enormous. The report warns that without patient capital and standardization, the “ChatGPT moment for robotics” may remain more aspiration than reality.

---

🧠 Tools to Try

Featured platforms and experimental agents for hands-on exploration.

Perplexity Model Council — Run Claude, GPT-5.2, and Gemini in parallel for cross-validated answers with reduced hallucination risk. https://www.perplexity.ai

2. ChatGPT Health — Connect your medical records and wellness apps for personalized health guidance inside an encrypted ChatGPT space. https://openai.com

3. Claude Code — Anthropic’s AI coding agent, now at $2.5B ARR, with multi-agent teams that can build complex projects autonomously. https://www.anthropic.com

4. DeepSeek V3.2 — Try DeepSeek’s latest publicly available model with expanded 1M-token context and updated knowledge cutoff (V4 launch imminent). https://www.deepseek.com

5. NVIDIA Physical AI Models — Open models on Hugging Face for robot learning and reasoning, enabling developers to build next-gen AI robots without training from scratch. https://huggingface.co/nvidia