5 May 2026: Trial Witness Warns of AGI Arms Race as Enterprise AI and SMB Tools Expand (AM)

The Musk v. OpenAI courtroom drama produced its most substantive AI safety testimony yet on Monday, as a leading academic researcher warned the world’s top AI labs are locked in a competition that could outpace humanity’s ability to govern it. At the same time, Anthropic and OpenAI announced new enterprise ventures, DoorDash handed small restaurants practical AI tools, and fresh data confirmed that image generation is now the biggest driver of AI app growth.

A top AI safety researcher testified at the Musk v. OpenAI trial this week that the AI industry risks an uncontrolled AGI arms race unless governments act. Stuart Russell, a professor at the University of California Berkeley and the author of the field’s most widely used textbook, took the stand as Elon Musk’s sole expert witness on Monday. Russell argued that without binding international oversight, frontier AI labs are caught in a competitive dynamic that treats safety as a cost rather than a priority.

His concern centres on artificial general intelligence, or AGI: AI that can perform any intellectual task a human can do. Russell told the court that governments need to introduce hard limits on what frontier labs can build and how fast, drawing comparisons to nuclear non-proliferation frameworks. The OpenAI trial itself turns on Musk’s claim that the company abandoned its original nonprofit safety mission when it accepted billions in commercial investment.

For UK readers, the stakes are direct. The UK government has invested heavily in positioning itself as a global AI safety leader, hosting the Bletchley Park Summit in 2023 and establishing the AI Safety Institute. Any legal ruling or regulatory momentum that reshapes how frontier labs operate in the US will influence how British regulators, businesses, and procurement teams approach AI systems in the months ahead.

Anthropic and OpenAI are both launching dedicated joint ventures to sell AI services directly into large enterprises, signalling a shift away from purely self-serve models. TechCrunch reported on Monday that both labs have each partnered with established asset managers and professional services firms to create co-owned businesses focused on accelerating AI adoption inside large organisations. For Anthropic, the venture extends its existing partnerships with AWS and Google Cloud; for OpenAI, it continues a pattern of high-profile commercial alliances.

For UK businesses evaluating AI suppliers, this matters practically. Both labs are moving toward bundled enterprise offerings that combine model access, consulting, and support. If you are scoping an AI deployment, expect more structured procurement options and dedicated account teams rather than a self-serve developer experience. The pace of enterprise deal-making from both companies suggests this shift is accelerating.

DoorDash has rolled out AI tools giving independent restaurant owners capabilities that previously required marketing agencies or professional photographers. The new features, launched Monday, let merchants use AI to automatically enhance food photography and speed up onboarding through AI-guided flows that pre-fill business information and suggest menu structures. Both features are available through the merchant portal and require no technical knowledge to use.

Better food photography directly affects how many customers click through to order, making this a genuinely useful tool for smaller operators. While DoorDash is a US company, similar features are likely to reach UK merchants as the platform rolls out updates across markets. The broader pattern is worth watching: AI is increasingly being embedded into existing small business platforms rather than sold as a separate product, lowering the barrier to adoption for businesses that have never called themselves tech companies.

New research from app analytics firm Appfigures shows that image generation AI features now drive 6.5 times more app downloads than chatbot or text AI upgrades. The data tracks what moves the needle when AI companies ship new capabilities, and the answer is clear: visual AI features consistently outperform language model improvements when it comes to acquiring new users. Image generation tools like those found in Adobe Firefly, Midjourney, and ChatGPT’s image features are winning the consumer acquisition battle.

The caveat is that image AI has so far struggled to convert download spikes into lasting revenue. For developers and businesses considering where to add AI features, the data points to image capabilities as the most effective hook for new user growth, even if the commercial case still needs to be proven.

Google has added event-driven Webhooks to the Gemini API, removing a key friction for developers building AI applications that handle time-consuming tasks. The update replaces a previous approach where applications had to repeatedly check whether a long-running AI job was complete. Now Google sends a push notification when the work is done. In practical terms, this means apps built on Gemini for tasks like generating long documents or processing large files no longer need clunky retry loops in their code. The change brings Gemini’s developer experience closer to modern cloud infrastructure patterns and makes it easier to build reliable production applications.

Here is everything else worth knowing from this morning’s AI news.

• Cerebras heading for a $26.6B IPO – The AI chip maker, which has a close commercial relationship with OpenAI, is on track for a blockbuster public listing. Cerebras builds chips designed specifically to run large AI models faster than conventional hardware. [4 May]
• Sierra raises $950M at $15B valuation – The enterprise AI platform, which builds AI agents for customer-facing interactions, has secured fresh capital the company says it will use to become the global standard for AI-powered customer experiences. [4 May]
• OpenAI and PwC team up on CFO tools – OpenAI announced a collaboration with PwC to bring AI into financial analysis, reporting, and forecasting workflows inside large organisations. [4 May]
• Nvidia’s Jensen Huang pushes back on AI job fears – Nvidia chief executive Jensen Huang argued publicly that AI is creating a large number of jobs rather than destroying them, at a moment when worker anxiety about AI-driven displacement is running high. [5 May]
• Google rounds up its April AI updates – Google published a summary of all its AI product announcements from April 2026, a useful single reference for anyone tracking where its consumer and developer tools stand heading into May. [4 May]

This is a daily news update for informational purposes only. AI products and policies change rapidly. Verify details directly with providers before making decisions. Nothing here is financial or legal advice.

AI Daily is Cristoniq’s daily guide to developments in artificial intelligence, published every morning.