Global AI News Roundup: Major Developments in Artificial Intelligence (May 12-18, 2025)

In a week marked by significant regulatory battles, technological advancements, and corporate realignments, the artificial intelligence landscape continues to evolve at a breathtaking pace. From government policy shifts to major AI product launches and academic controversies, here’s a comprehensive look at the most important global AI news developments from May 12 to May 18, 2025.

US Government Reverses AI Chip Export Ban Amid Industry Pressure

The Biden-era “AI diffusion rule,” which aimed to restrict AI chip exports to over 100 countries, has been officially scrapped just days before its May 15 implementation deadline. The controversial policy faced significant pushback from major tech companies that argued it was overly broad and counterproductive.

The U.S. government now plans to pursue more targeted bilateral agreements to limit chip transfers, particularly those that could benefit China’s military AI initiatives. This strategic shift aims to maintain America’s competitive edge while avoiding harm to allies and domestic industry AI Security Weekly.

Republicans Propose 10-Year Ban on State AI Regulations

In a controversial move, Republicans in the US Congress have introduced a provision in a proposed budgetary bill that would prohibit states from implementing or enforcing laws regulating artificial intelligence or automated decision-making systems for the next decade.

The last-minute addition to the bill would effectively bar state and local governments from creating guardrails for AI systems unless the purpose is to remove legal impediments to or facilitate the deployment of these technologies. This could potentially invalidate existing state laws, such as New York’s requirement for automated hiring systems to undergo bias assessments and California’s regulations on AI in healthcare communications The Guardian.

Critics argue this provision serves to shield powerful corporations from accountability, with Lee Hepner from the American Economic Liberties Project describing it as a “sweeping and reckless attempt” that would benefit tech giants at the expense of consumer protections.

Google One Reaches 150 Million Subscribers, Boosted by AI Features

Google’s subscription service has achieved a significant milestone, surpassing 150 million subscribers—a 50% increase since February 2024. The growth has been substantially driven by the company’s new $19.99 monthly AI-focused tier, which has already attracted “millions” of subscribers since its introduction earlier this year.

This subscription growth is part of Alphabet’s broader strategy to diversify revenue beyond advertising as the company navigates the potential disruption AI chatbots pose to its core search business. Google CEO Sundar Pichai has emphasized the subscription direction as a key monetization strategy for the company’s AI offerings in 2025 Reuters.

Amazon’s Alexa+ AI Launch Faces Adoption Challenges

Six weeks after Amazon began rolling out its AI-powered Alexa+ voice assistant to what it claims are “hundreds of thousands” of users, there appears to be little evidence of widespread adoption. Despite CEO Andy Jassy’s assertion that “people are really liking Alexa+ thus far,” technology analysts and media outlets have struggled to find verifiable users or reviews of the service.

Sources familiar with the matter indicate that the rollout is proceeding slowly due to challenges with response speed, occasional inaccuracies, and high operational costs. The situation highlights the difficulties tech companies face when transitioning from traditional services to more complex AI-powered offerings Reuters.

MIT Disavows AI Productivity Research Paper Due to Data Integrity Concerns

In a significant blow to AI productivity research, MIT has requested the withdrawal of a high-profile paper that claimed to demonstrate AI tools boosting scientific discovery and patent filings in materials science. The university cited serious concerns about “the integrity” of the research and stated it has “no confidence in the provenance, reliability or validity of the data.”

The paper, which had been praised by notable economists including Nobel laureate Daron Acemoglu, was scheduled for publication in The Quarterly Journal of Economics. This controversy underscores growing concerns about research quality and verification standards in the rapidly expanding field of AI studies TechCrunch.

Elton John Condemns UK Government’s AI Copyright Proposals

Renowned musician Sir Elton John has strongly criticized the UK government over its proposals regarding AI and copyright law, calling officials “absolute losers” for considering allowing tech firms to use copyright-protected work without permission. Speaking on BBC’s Sunday with Laura Kuenssberg program, John argued the government was effectively “committing theft” and would “rob young people of their legacy and income.”

The controversy centers around the government’s consultation on allowing AI firms to train their models on copyrighted work without explicit permission, using an opt-out rather than opt-in approach. This comes as the House of Lords considers an amendment that would require AI companies to disclose their use of copyright-protected content The Guardian.

AI Companies Report Major Business Developments

Several AI companies announced significant business developments this week:

• Cohere: The AI startup has doubled its annualized revenue to $100 million, with growth driven by demand for secure, customized AI solutions from enterprises in regulated sectors. A strategic pivot toward private deployments and tailored models now represents 85% of its business LinkedIn.
• Microsoft: The company laid off 2,000 employees in Washington state—over 40% were software engineers—as part of a broader 6,000-position reduction. This occurred after CEO Satya Nadella noted AI now writes up to 30% of the company’s code LinkedIn.
• Windsurf: The company launched its own family of AI models, SWE-1, designed specifically for software engineering workflows. Reports suggest OpenAI considered acquiring Windsurf for approximately $3 billion, highlighting the strategic value of specialized AI systems LinkedIn.

UAE-US Partnership Establishes Massive AI Computing Infrastructure

The UAE and US have signed an agreement to build a 5 GW-capacity AI campus in Abu Dhabi—one of the largest computing facilities outside the United States—with strict security controls and an initial 1 GW data center. Separately, OpenAI is reportedly in discussions with UAE firms G42 and MGX regarding additional data center projects in the region.

In Italy, Abu Dhabi’s G42 AI group and Italian company iGenius have announced plans to construct what they describe as Europe’s largest AI computing infrastructure, likely to be located in Apulia. G42 will lead financing for the initial supercomputer phase, representing a major advancement in Europe’s AI capabilities through international collaboration LinkedIn.

Google Expands AI Mode in Search to All US Users

Google has removed the waitlist for its AI Mode in Search, making it available to all users in the United States through Google Labs. The company is also introducing new features to help users complete tasks more efficiently, including visual place and product cards with detailed information.

For local businesses like restaurants and stores, AI Mode will now display information such as ratings, reviews, and opening hours, while product searches will include real-time prices, promotions, and inventory information. Google also introduced a new left-side panel for desktop users that preserves search history by topic Google Blog.

Security Concerns Highlighted at RSA Conference 2025

The RSA Conference spotlighted growing dangers associated with autonomous AI agents, from identity impersonation to unintended data exposure. Security vendors and researchers emphasized the urgent need to build protective measures into agent lifecycles before widespread deployment.

As AI agents increasingly act on behalf of users and organizations, experts warn that without rigorous security controls—particularly in identity and permissions management—these systems could create powerful new attack surfaces. The consensus from the conference suggests agent security must become a core pillar of AI architecture AI Security Weekly.

Conclusion: AI’s Expanding Influence Amid Growing Regulatory Questions

This week’s developments highlight the accelerating integration of AI into business operations, government policy, and creative industries. As technological capabilities advance, questions about regulation, copyright, job displacement, and security grow increasingly urgent.

The contrasting approaches to AI governance—from the US Republicans’ proposed deregulation to the UK’s ongoing copyright battles—underscore the complex balancing act between fostering innovation and protecting established rights and industries. Meanwhile, tech giants continue to refine their AI business models, with subscription services emerging as a preferred monetization strategy.

As AI capabilities continue to evolve rapidly, the need for thoughtful policy frameworks that can keep pace with technological change remains one of the most significant challenges facing governments, businesses, and society at large.