Daily AI Brief

JPMorgan's Vice Chairman Daniel Pinto cautioned that many AI-linked stocks might be over-valued and could face a correction as expectations collide with the realities of monetisation. The warning forces a re-examination of the "growth-at-all-cost" narrative in the AI industry.

Sundar Pichai stated in an interview that AI models remain prone to error and should be used alongside other tools — he also warned no company is immune if the AI bubble bursts. The remarks underline how even major AI-players recognise the risk of over-reliance on today's systems.

An app called 2wai allows users to create avatars of dead relatives from a few minutes of video and converse with them via AI; critics condemned it as "vile" and dystopian. The story highlights urgent ethical questions in generative-AI personalisation and digital-legacy tech.

A new report reveals that while two-thirds of US health-systems have multi-person governance committees, more than 75% rely on just two or fewer full-time staff for oversight — and AI project requests out-pace governance decisions nearly 3:2. The gap suggests deployment risks are growing even as AI adoption increases.

The ethics and oversight of AI in creative and cultural domains are now front-of-mind, as evidenced by the New Zealand book-prize decision treating AI-designed covers as disqualifying. The precedent signals that 'AI-used' will increasingly require disclosure or come with limitations.

Meanwhile, major industry voices (e.g., Alphabet's CEO) are raising alarms about tool trustworthiness and oversight gaps, feeding into regulatory debates around transparency, reliability and accountability of AI deployment – especially in mission-critical sectors.

The startup-funding story of over US$1.5 billion raised by a cloud AI-infrastructure company (Lambda) underscores how capital is flowing into the backbone of the AI ecosystem, not just models.

This Guardian long-form piece explores how reliance on generative AI may erode traditional wisdom and diverse epistemologies, raising questions about what is lost as much as what is gained.

"If you don't call out the risks now, you'll be stuck responding to the fallout later." — Dario Amodei, CEO of Anthropic (in context of broader risk commentary).

As highlighted by today's governance-survey story, this week's focal tool is governance architecture (policies, staffing, oversight workflows) rather than a specific software product, signalling that successful AI deployment equals management tool, not just model-tool.