Global AI News Daily — 2026-04-28

Top Stories

China Orders Meta to Unwind Acquisition of AI Startup Manus

China has required Meta to unwind its acquisition of AI startup Manus, according to a New York Times report published April 27. The ruling's full impact remains unclear, but it sends a strong signal to Chinese tech founders weighing partnerships with foreign companies. The move underscores growing geopolitical friction over cross-border AI investments and could significantly dampen future collaboration between Chinese AI talent and Western tech giants.

DeepMind's David Silver Raises $1.1B for AI That Learns Without Human Data

Ineffable Intelligence, a British AI lab founded just months ago by former DeepMind researcher David Silver — the mind behind AlphaGo — has raised $1.1 billion at a $5.1 billion valuation. The funding, reported by TechCrunch just hours ago, will fuel the development of novel AI models designed to learn without human-generated data, a fundamentally different approach from today's dominant large language models. The raise signals investor appetite for next-generation AI architectures that could leapfrog current scaling paradigms.

AI Companies Compete With "Bragawatts" — Massive, Unverified Energy Claims

A New York Times analysis published April 26 coins a new term for AI companies' habit of boasting about how much energy capacity they have access to — often without evidence: "bragawatts." The piece explores how exaggerating energy access has become a competitive and intimidation tactic in the AI industry, raising questions about transparency and the actual state of AI infrastructure buildout. As data center demand soars, the gap between claimed and actual capacity may be widening.

Company Watch

OpenAI CEO Sam Altman Warns AI Could "Collapse" the Economy — OpenAI CEO Sam Altman posted on X that he is "literally losing sleep" over the potential for AI technology to collapse the economy, according to a Times of India report published April 27. Altman's comments add to a string of high-profile warnings about AI's macroeconomic disruption potential, coming from the very person leading one of the world's most powerful AI labs.

Frontier AI Model Comparisons Heat Up — GPT-5.5, Claude Opus 4.7, Gemini 3.1 Pro, DeepSeek V4 — A BuildFastWithAI roundup published two days ago benchmarked eight frontier models across coding, research, and chat tasks. The report highlights that DeepSeek V4 costs approximately 85% less than GPT-5.5, making cost-performance tradeoffs a central battleground for enterprise AI adoption in 2026.

AI Stocks Approach All-Time Highs as Nasdaq Rebounds — With the Nasdaq approaching record territory, Motley Fool analysis published April 26–27 flags select AI growth stocks as potential long-term winners. One stock highlighted is up more than 300% year-to-date, reflecting sustained investor confidence in AI infrastructure and application plays despite broader market volatility.

Policy & Regulation

California Governs Generative AI Procurement Across State Government — On March 30, 2026, California Governor Gavin Newsom signed executive order N-5-26, aimed at governing responsible procurement and deployment of generative AI across California's government agencies. According to Alston & Bird's April 2026 AI Quarterly (published within the past week), the order is part of a broader wave of U.S. state-level AI governance efforts moving in parallel with — and sometimes ahead of — federal action.

EU AI Act Implementation Support Tools Coming in Q2 2026 — The European Commission confirmed that support instruments to assist organizations in complying with the EU AI Act are under preparation and will be published in the second quarter of 2026. This means businesses operating in Europe face an imminent wave of new compliance guidance, even as debate continues about whether some provisions should be paused. The EU's approach continues to contrast sharply with the more permissive stance emerging in parts of the U.S.

Industry Moves

Ineffable Intelligence Closes $1.1B Series A at $5.1B Valuation — The David Silver-founded British AI startup Ineffable Intelligence has officially closed a $1.1 billion funding round, vaulting to a $5.1 billion valuation just months after its founding. The lab is pursuing AI architectures capable of learning without human data — a bet on a fundamentally different path to superintelligence than today's transformer-based systems.

Meta's Manus Acquisition Faces Forced Divestiture — China's regulatory action requiring Meta to unwind its Manus acquisition represents one of the most significant cross-border AI deal interventions to date. The move could force Meta to reassess its international AI acquisition strategy and may chill deal-making between Western tech companies and Chinese AI startups more broadly.

AI Model Benchmarking Becomes a Competitive Industry — Analyst and developer-facing platforms such as BuildFastWithAI and GuruSup are now publishing detailed head-to-head comparisons of frontier models across GPT-5.5, Claude Opus 4.7, Gemini 3.1 Pro, and DeepSeek V4. The emergence of structured model benchmarking as a genre reflects the maturing enterprise AI market, where procurement decisions increasingly hinge on performance-per-dollar metrics.

What to Watch

1. EU AI Act Compliance Guidance Drop (Q2 2026): The European Commission has signaled that implementation support tools for the EU AI Act are imminent — expect formal publication in the coming weeks. Organizations operating in Europe should prepare for a compliance sprint.

2. Ineffable Intelligence Research Direction: With $1.1 billion secured, David Silver's lab will soon begin publishing details on its approach to human-data-free AI learning. Early research disclosures or hiring announcements could give the first real signal of whether this approach can challenge current frontier models.

3. Meta's Manus Divestiture Timeline: Following China's ruling, observers are watching for Meta's official response and how quickly regulators will enforce the unwind. The outcome could set a precedent for future cross-border AI M&A involving Chinese-founded startups.

Quick Reads

"Bragawatts": The New AI Arms Race Is About Energy Bragging Rights — NYT coins a term for AI companies exaggerating energy access to intimidate rivals and attract investors, often without evidence.

DeepSeek V4 Costs 85% Less Than GPT-5.5 — A Mashable comparison of frontier models highlights DeepSeek's dramatic cost advantage, raising pressure on OpenAI's enterprise pricing strategy.

Beyond the Hype: The Most Interesting AI Models Aren't the Famous Ones — XDA Developers argues that the most exciting AI innovations in 2026 are coming from labs and models that don't dominate mainstream headlines.

Sam Altman Says AI Will "Collapse" the Economy — And He Can't Sleep — The OpenAI CEO's latest warning about AI's economic disruption is generating significant attention.

California Now Governs How Its Own Government Buys AI — Governor Newsom's executive order N-5-26 makes California one of the first U.S. states to formally regulate government AI procurement.