AI Daily Briefing — 2026-05-16

China's short drama industry is undergoing a radical transformation, with artificial intelligence now producing hundreds of shows daily. In January alone, an average of 470 AI-generated dramas were released, slashing production costs by up to 90% and shrinking timelines from months to weeks. These bite-sized, melodramatic shows—once reliant on human actors and crews—are now created entirely by algorithms, with storytelling increasingly guided by performance data. The format is expanding globally, reshaping the work of writers and production teams, as reported by Caiwei Chen for MIT Technology Review AI.

Meanwhile, the World Health Organization's latest global statistics report paints a grim picture of stalled progress on major health threats. New HIV cases reached 1.3 million in 2024, malaria is resurging, vaccination rates are declining in the Americas, and 42.8 million children face severe malnutrition. The world is significantly off track to meet the UN's 2030 health goals, according to Jessica Hamzelou in MIT Technology Review AI's The Checkup newsletter.

In other tech news, Elon Musk and Sam Altman face accusations of lying as their trial heads to the jury, with lawyers questioning their credibility. AI data centers are straining power grids, with Nevada redirecting electricity from Lake Tahoe to support the technology. OpenAI is considering legal action against Apple over its ChatGPT integration, while Anthropic has agreed to a $30 billion funding deal at a $900 billion valuation. Washington and Beijing are set to hold formal talks on AI safety, addressing guardrails and preventing nonstate actors from accessing powerful models.

A new wave of ultra-short, melodramatic series is flooding smartphone screens, but many are now made entirely by artificial intelligence. One example, *Carrying the Dragon King’s Baby*, features a woman thrown onto a bed by a muscular man, with flame-like vines crawling across her body. The glossy but slightly uncanny visuals—somewhere between a movie and a video game cutscene—reveal the secret: no actors, cameras, or CGI specialists were involved. Instead, AI generated the entire production.

China’s short drama industry exploded after launching in 2018, with episodes lasting just one to two minutes. Designed for endless scrolling, these shows rely on cliffhanger-heavy ads on TikTok and Instagram to lure viewers into paid subscriptions. By 2024, the market hit roughly $6.9 billion in revenue, surpassing China’s annual box office. Globally, short drama apps have approached a billion downloads, with the U.S. providing about half of overseas revenue.

Now the industry is reinventing itself with generative AI. In January alone, an average of 470 AI-generated short dramas were released daily, according to research firm DataEye. Production timelines have collapsed from three to four months to under a month, and costs have dropped by 80 to 90 percent. Companies like Kunlun Tech are shrinking film crews and reorganizing labor pipelines, with AI moving from a supporting tool to the backbone of production.

Decisions about what to produce are driven by performance data rather than creative instinct. Platforms categorize themes and plotlines that resonate, then quickly adjust. “In China, if a series doesn’t break even within a month, it’s considered a failure,” said Tang Tang, vice president of FlexTV. The relentless pace means screenwriters now work within tight algorithmic constraints, churning out content optimized for viral consumption.

When Jennifer, a researcher, ran her professional headshot through facial recognition in 2023, she expected to find old porn videos from her past. Instead, she discovered something far more disturbing: her body in an explicit video, now wearing a stranger's face. While public outrage over deepfake porn typically centers on victims whose faces are stolen, adult content creators say they are being erased from their own work. AI systems trained on their videos are cloning their bodies to generate new explicit material without consent, leaving them with little legal recourse. MIT Technology Review AI reports these performers face a unique crisis—their physical likenesses are being exploited while their rights to their own bodies are ignored.

Meanwhile, a separate privacy crisis is unfolding as AI chatbots leak personal contact information. A software developer received unsolicited WhatsApp messages after Google's Gemini surfaced his number. A university researcher tricked the same chatbot into revealing a colleague's private cell phone. Experts trace these breaches to personally identifiable information embedded in training data, which chatbots now make easily searchable. Victims have few options to stop the leaks, raising urgent questions about accountability in AI deployment.

In other tech news, Tesla's long-awaited Semi truck is finally rolling off production lines, promising to revolutionize electric freight. With a range of 480 miles per charge and lower costs than competitors, the Semi could slash pollution from semitrucks—a major source of transport emissions. MIT Technology Review AI notes this breakthrough comes as the US approved Nvidia chip sales to Chinese firms including Alibaba and Tencent, while Beijing pushes for domestic AI independence. These developments highlight the complex interplay between innovation, regulation, and global power dynamics.

A nine-person jury in California is now deliberating the fate of OpenAI, the prominent AI lab, in a case brought by Elon Musk against co-founders Sam Altman and Greg Brockman, along with Microsoft. The trial has touched on the founders' 2018 split and Altman's 2023 ouster and return, but the jury must answer narrow legal questions. They will decide if the defendants breached a charitable trust by misusing Musk's donations, if they were unjustly enriched through OpenAI's for-profit arm, and if Microsoft aided that breach.

OpenAI's defense rests on three arguments. They claim the statute of limitations bars Musk's claims, as any alleged harm occurred before key dates in 2021 and 2022. They also argue Musk unreasonably delayed his 2024 lawsuit, and that his own conduct—including attempts to merge OpenAI with Tesla—amounts to "unclean hands," invalidating his case. A forensic accountant testified that Musk's donations were fully spent before August 2021, undermining his charitable trust claim.

If Musk prevails, OpenAI could be forced to unwind its for-profit structure, though the exact consequences remain unclear. The judge will hold hearings next week to debate potential remedies if the verdict favors Musk. A loss for the plaintiff, however, would render those proceedings moot. The case hinges on whether Musk's donations were explicitly tied to a non-profit mission that was later abandoned for profit.

Richard Socher, the founder of You.com and a veteran of ImageNet, has launched a new San Francisco-based startup called Recursive Superintelligence. The company emerged from stealth on Wednesday with $650 million in funding, backed by a team of prominent AI researchers including Peter Norvig and Cresta co-founder Tim Shi. Their mission is to create a recursively self-improving AI model that can autonomously identify its own flaws and redesign itself without human input—a long-sought goal in the field.

In an interview after the launch, Socher explained that the startup's unique approach relies on open-endedness, a concept that allows AI systems to evolve continuously by competing and adapting, much like biological evolution. He cited examples such as co-founder Tim Rocktäschel's work on Genie 3, a world model that generates interactive environments, and a technique called rainbow teaming, where two AIs challenge each other to find vulnerabilities, leading to safer systems over millions of iterations.

Socher emphasized that true recursive self-improvement goes beyond simple automation. It requires the entire cycle of ideation, implementation, and validation of research ideas to be automatic, starting with AI research and eventually extending to physical domains. He described this as a form of self-awareness, where the AI recognizes its own shortcomings and works to fix them. The startup aims to achieve this at scale, a feat that has so far eluded other labs.

The company joins a wave of research-focused AI startups that prioritize breakthroughs over products. Socher, however, does not see Recursive Superintelligence as a typical neolab. Instead, he views it as a step toward building a superintelligence that can improve itself indefinitely, a vision that has captivated AI researchers for decades.