Anthropic Claude Outage Cripples AI, Meta's Data Leak Scandal, MGM Shelves Sam Altman Biopic, China Tops Supercomputer Race!

**Adam Nagus** (0:05)
Welcome to Digimasters Shorts. We are your hosts, Adam Nagus.

**Carly Wilson** (0:09)
And Carly Wilson, delivering the latest scoop from the digital realm.
Claude is back online after a major outage disrupted service for most users across most of its models. Anthropic acknowledged the issue and said it deployed a fix as of 10:53 a.m. Eastern time after the disruption began around 10:19 a.m. The outage affected multiple models, though the company did not specify which ones were involved. Attempts to access popular models such as Sonnet and Opus during the disruption returned no responses. The incident impacted platforms including claude.ai, Claude Console, Claude Code and the Claude API, but did not affect Claude for government. The downtime lasted nearly an hour, making it one of the most significant outages for Anthropic in the past two months. Fable 5 had already been removed from user access prior to the disruption. Anthropic has not explained the cause of the outage, but says it is monitoring the system as service returns to normal. For now, Claude appears to be operating as usual.

**Adam Nagus** (1:04)
Meta has paused an internal data tracking program after a leak exposed sensitive employee information across the company. Reports say the breach made full AI prompts, transcriptions, performance metrics, and even private conversations accessible to other employees. The initiative, known as the Model Capability Initiative, was designed to help Meta's AI systems learn how people complete everyday computer tasks. It reportedly collected extensive data, including keystrokes and screen recordings. The program sparked immediate backlash when it was introduced in April, with employees calling it an invasion of privacy and circulating a petition to shut it down. The controversy unfolded amid low morale following nearly 8,000 layoffs and increased pressure from leadership to accelerate AI development. Chief Technology Officer Andrew Bosworth had previously assured staff that any collected data would be tightly controlled, but later acknowledged the rollout fell short of privacy standards. Meta says it has found no evidence of malicious access, but is investigating and will not reinstate the program until stronger safeguards are in place. The company also revealed that an initial attempt to fix the security flaw failed before the issue was ultimately resolved. The incident marks another security setback for Meta, following a recent episode in which a rogue AI agent exposed sensitive information in what was described as a critical breach.
An unflattering biopic about OpenAI CEO Sam Altman has reportedly been dropped by Amazon-owned MGM Studios despite being nearly complete. The film, titled Artificial and directed by Luca Guadagnino, was shelved with Amazon saying it would be better served by a different studio. The decision comes months after Amazon pledged $50 billion to OpenAI in a funding round that pushed the AI company's valuation above $700 billion.
OpenAI also agreed to purchase $38 billion in cloud services from Amazon, underscoring deep financial ties between the two firms. Reports suggest the film portrays Altman as a pathological liar, drawing comparisons to the social network rather than offering a flattering depiction.
Test screenings were said to be positive and Andrew Garfield reportedly plays Altman. Observers question whether the shelving reflects growing influence from powerful tech executives over major media companies. The episode highlights increasingly intertwined relationships between the tech and entertainment industries, particularly as AI investment accelerates. With Netflix focus features and A24 all passing on distribution, the film's prospects appear uncertain. The controversy adds to concerns about a tech elite seen as closing ranks to protect its most prominent figures.

**Carly Wilson** (3:40)
China's Lineshine supercomputer has taken the top spot in the latest top 500 rankings, marking the first time since 2017 that a Chinese system has led the list, often seen as a barometer of national technological strength.
Making its debut, Lineshine displaced the US machine El Capitan, which now ranks second. Housed at the National Supercomputing Center in Shenzhen, Lineshine achieved 2.198 exaflops, meaning it can perform more than two quintillion calculations per second. Unlike many high-performance systems optimized for artificial intelligence, it runs entirely on conventional CPUs rather than GPUs, and consumes about 42.2 megawatts of electricity.
Supercomputers of this scale are used for tasks ranging from medical research and climate modeling to nuclear simulations and virtual weapons testing. El Capitan, located at Lawrence Livermore National Laboratory in California, now leads a group of three US systems in the top four, while Germany's Jupiter ranks fifth. These five machines are currently the only publicly verified exascale supercomputers in operation worldwide. The United Kingdom has 11 systems on the list, led by the University of Bristol's Isambard AI, while Australia's highest ranked machine, Setanix, sits at 86th. Meanwhile, the European Union has announced a $20 billion plan to develop AI gigafactories powered by massive supercomputers, aiming to compete with leaders in the United States and China, while relying as much as possible on green energy.