🗞 OpenAI Returns to Nonprofit Roots, Google Tests AI Controls, and Anthropic Launches Science Program

🔎 The Latest on the AI Frontier:

• OpenAI Pivots Back to Nonprofit Control: Company abandons for-profit conversion plans, establishing unique structure with nonprofit oversight

• Google Tests Computer Use Tools for AI: Hidden code reveals development of AI agents that can control screens and applications

• Anthropic Launches Science Research Program: New initiative offers researchers API credits to access Claude models for scientific advancement

• Tech Job Market Contracts Despite Economic Growth: IT positions decline while AI expertise becomes increasingly valuable in shifting landscape

• Uber Teams With China's Pony AI for Middle East Robotaxis: Partnership aims to deploy autonomous vehicles in ride-hailing service by year's end

• Other news you might find interesting

🔄 OpenAI reverses course on corporate structure, will maintain nonprofit control

• OpenAI Chair Bret Taylor announced the company is abandoning plans to convert to a for-profit entity and will instead transition its business arm to a Public Benefit Corporation (PBC) while keeping it under nonprofit oversight, a decision made after "constructive dialogue" with attorneys general from Delaware and California .

• The move comes with financial implications, as some investors who provided $6.6 billion on the condition of a for-profit conversion will either have the right to reclaim their capital or receive a 9% interest rate if the conversion doesn't occur within two years—potentially costing the company over $600 million .

• CEO Sam Altman emphasized in an internal letter that "OpenAI is not a normal company and never will be," with the restructuring giving employees and investors equity while the nonprofit becomes a large shareholder in the PBC—an unprecedented arrangement for a company reportedly valued at nearly $300 billion .

🛠️ Google quietly tests Computer Use tools and Cloud Run hosting in AI Studio

• Evidence briefly appeared in Google AI Studio's web version showing a new "Computer Use" capability - industry terminology for letting AI agents control screen, cursor, and native applications - before being quickly removed from the code.

• Google's approach appears to combine this agent control framework (researched under the "Jarvis initiative") with simplified Cloud Run deployment for Gemma 3 models, potentially allowing developers to create containerized instances directly from the browser interface.

• The integration would streamline development by connecting prompt design directly to deployment, allowing tasks to run on-device while handling heavier processing in the cloud - though whether this becomes a public feature likely depends on safety evaluations and Google's broader agent strategy.

🔬 Anthropic introduces AI for Science program to accelerate scientific research

• The program offers selected researchers up to $20,000 in API credits over six months, providing access to Anthropic's standard AI model suite including all publicly available Claude models.

• Applications will be evaluated on scientific merit, potential impact, technical feasibility, and biosecurity screening criteria, with selections made on the first Monday of each month by Anthropic's team including subject matter experts.

• Despite industry enthusiasm for AI in scientific discovery, many researchers remain skeptical of current AI tools' usefulness, as previous AI science initiatives have shown mixed results—like Google's GNoME, which claimed to help synthesize 40 new materials that an outside analysis found weren't actually new.

💻 US tech job market shrinks despite overall employment growth, with AI expertise increasingly in demand

• The US economy added 39,000 more jobs than expected in April, but IT consultancy Janco's analysis found 10,600 IT jobs disappeared that month after the same quantity of losses in March, while tech certification org CompTIA concluded 214,000 tech jobs overall vanished.

• The downward trend may be driven by multiple factors, including increased automation as "companies do not have the desire to hire new staff to meet mandated compliance requirements" and are "focusing on AI to automate as many of those tasks as possible," according to Janco CEO Victor Janulaitis.

• Other contributing factors include the Trump administration's policies and the effects of Elon Musk's cost-trimming "DOGE team," with outsourcing companies "scrambling to maintain existing contracts" as businesses prepare for tariff impacts, though AI implementations and security roles remain bright spots, with AI-related job postings increasing 184% last month.

🚕 Uber partners with China's Pony AI to deploy self-driving taxis in Middle East market

• Uber announced a partnership with China-based robotaxi developer Pony AI to integrate autonomous vehicles into its ride-hailing platform, beginning with an initial pilot featuring safety operators before transitioning to fully autonomous commercial service in a key Middle East market later this year .

• The collaboration represents Uber's latest strategic move in the autonomous vehicle sector, following recent partnerships with May Mobility and Momenta, an expanded agreement with China's WeRide to enter 15 more cities globally, and last year's extension of its tie-up with Alphabet's Waymo .

• Founded in 2016 and backed by Toyota, Guangzhou-based Pony AI currently holds robotaxi service licenses in Beijing, Shanghai, Guangzhou and Shenzhen, with its U.S.-listed shares jumping nearly 13% in premarket trading following the announcement.

More news you might find interesting:

• Anthropic launches 'Beneficial Deployments' team focused on AI for social good, led by former Biden official.

• Visa launches Intelligent Commerce platform enabling AI agents to make purchases on users' behalf.

• Researchers develop MagicTime AI model capable of generating realistic metamorphic videos.

• Disney employee falls victim to malicious AI tool scam resulting in massive 1.1TB data theft.

• The growing resistance: meet the people refusing to use AI despite its widespread adoption.

• Neuralink's first nonverbal recipient demonstrates brain chip capabilities with self-edited video.

Have any feedback? Send us an email