philip lelyveld The world of entertainment technology

China to regulate use of digital humans

...

The draft requires that providers of services involving digital humans, which are human-like virtual beings capable of voice chat or other forms of interaction, must continuously display a prominent label reading "digital human" alongside such content.

According to the draft, the use of individuals' sensitive personal information for generating and modeling digital human images must be subject to their informed consent.

It stipulates that no organization or individual may use digital humans with identifiable traits of specific natural persons without their consent.

...

See the full story here: https://english.news.cn/20260403/ae0d10c6c7d4408e8f7c407593bbb1b2/c.html

Penalties stack up as AI spreads through the legal system

...

In the meantime, AI itself has strayed into the crosshairs of the legal profession. In March, OpenAI, the maker of ChatGPT, was sued by Nippon Life Insurance Company of America in federal court in Illinois. The insurance company says it was the target of frivolous legal actions by a woman who was getting bad legal advice from ChatGPT.

Among other things, the lawsuit accuses OpenAI of practicing law without a license. In a written statement to NPR, OpenAI responded, "This complaint lacks any merit whatsoever."

See the full story here: https://www.npr.org/2026/04/03/nx-s1-5761454/penalties-stack-up-ai-spreads-through-legal-system

Manatt Launches AI Consulting and Testing Service Offering

Manatt announced the launch of its new AI Consulting and Testing (AICT) service offering, a multidisciplinary initiative that enhances Manatt’s existing legal and consulting experience with new testing capabilities to help organizations safely and responsibly build, deploy and oversee AI systems.

This new service offering includes AI testing, regulatory‑aligned governance, model validation and outcome-based testing, including bias testing, procurement support and ongoing monitoring—giving clients a defensible, end‑to‑end approach to responsible AI. ...

AICT supports clients that are developing or deploying AI tools across highly regulated and fast‑moving sectors—including health care, financial services, consumer products, entertainment and employer AI use cases—offering services across four pillars of responsible AI: Strategy & Governance; Testing & Evaluation; Procurement & Vendor Oversight; and Ongoing Monitoring & Compliance.  ...

see the full story here: https://www.manatt.com/insights/news/2026/manatt-launches-ai-consulting-and-testing-service-offering

AI benchmarks are broken. Here’s what we need instead.

...

It soon became clear that the benchmark tests on which medical AI models are assessed do not capture how medical decisions are actually made. Hospitals rely on multidisciplinary teams—radiologists, oncologists, physicists, nurses—who jointly review patients. Treatment planning rarely hinges on a static decision; it evolves as new information emerges over days or weeks. Decisions often arise through constructive debate and trade-offs between professional standards, patient preferences, and the shared goal of long-term patient well-being. No wonder even highly scored AI models struggle to deliver the promised performance once they encounter the complex, collaborative processes of real clinical care. ...

When high benchmark scores fail to translate into real-world performance, even the most highly scored AI is soon abandoned to what I call the “AI graveyard.” ...

HAIC benchmarks reframe current benchmarking in four ways: 

1.     From individual and single-task performance to team and workflow performance (shifting the unit of analysis)

2.     From one-off testing with right/wrong answers to long-term impacts (expanding the time horizon)

3.     From correctness and speed to organizational outcomes, coordination quality, and error detectability (expanding outcome measures)

4.     From isolated outputs to upstream and downstream consequences (system effects)

...

See the full story here: https://www.technologyreview.com/2026/03/31/1134833/ai-benchmarks-are-broken-heres-what-we-need-instead/

Nvidia CEO Jensen Huang’s advice to workers scared of AI: You’re just confusing your job with the tools you use to do it

...

However, Huang’s not convinced that AGI will cause a jobs reckoning, just like previous tech transformations weren’t able to snatch his job. 

“I’m the longest-running tech CEO in the world: 34 years,” he continued. “The tools that I’ve used to do my job have changed continuously in the last 34 years, and sometimes quite dramatically.”

Huang explained his reasoning through a job paradox he’s witnessed in the era of AI: the fate of radiologists. 

The Nvidia leader said that computer scientists predicted that the first job to be automated by AI would be radiology, since computer vision was going to reach “superhuman levels” thanks to the advanced tech. Around 2020 the tools became that powerful, Huang pointed out, and “every radiology platform and package today is driven by AI”—yet the number of radiologists has grown. ...

See the full story here: https://fortune.com/2026/04/01/nvidia-ceo-jensen-huang-advice-workers-scared-ai-confusing-job-with-tools-to-do-it/

What to Know About California’s Executive Order on A.I.

...

California has been a leader in tech lawmaking, and was the first state to pass a law mandating safety and transparency from the biggest A.I. companies. Mr. Newsom, a Democrat, signed the order partly as a message to President Trump, who has been trying to bat down state attempts to regulate A.I. ...

If the federal government designates a company a supply chain risk, which the Pentagon has recently done with the A.I. start-up Anthropic, California will conduct its own assessment. If the company isn’t determined to be a risk, the state may allow it to remain a contractor. ...

See the full story here: https://www.nytimes.com/2026/03/30/technology/california-ai-executive-order.html

AI Is Rewriting the Old Rules of Google Search and SEO

Winning the search war now depends less on keywords and more on what strangers are saying about you on Reddit...

See the full story here: https://www.wsj.com/tech/ai/google-search-seo-ai-tactics-5efb9f99

Pentagon to adopt Palantir AI as core US military system, memo says

• Letter emphasizes Palantir's AI as cornerstone of Pentagon strategy
• Palantir's Maven uses AI to help identify targets for strikes
• Making Maven program of record would secure long-term funding for Palantir

See the full story here: https://www.reuters.com/technology/pentagon-adopt-palantir-ai-as-core-us-military-system-memo-says-2026-03-20

What’s the right path for AI?

....

By contrast, Hao offered, an alternate path for AI might exist in the example of AlphaFold, the Nobel Prize-winning tool used to identify protein structures. This represents the concept of the “small, task-specific AI model tackling a well-scoped problem that lends itself to the computational strengths of AI,” Hao said.

She added: “It’s trained on highly curated data sets that only have to do with the problem at hand: protein folding and amino acid sequences. … There’s no need for fast supercomputing because the datasets are small, the model is small, and it’s still unlocking enormous benefit.” ...

See the full story here: https://news.mit.edu/2026/right-path-for-ai-karen-hao-paola-ricaurte-0320

Study finds ChatGPT gets science wrong more often than you think

A new study put ChatGPT to the test by asking it to judge whether hundreds of scientific hypotheses were true or false—and the results were far from reassuring. While the AI got it right about 80% of the time on the surface, its performance dropped significantly when accounting for random guessing, revealing only modest reasoning ability. Even more concerning, it frequently contradicted itself when asked the exact same question multiple times, sometimes flipping answers back and forth.

See the full story here: https://www.sciencedaily.com/releases/2026/03/260317064452.htm