US political rift and eyed Ohio seat, driverless trucks and Meta’s AI return, House of the Dragon wraps
Developed over time first seen 2 months ago
Political pressure around artificial intelligence intensified in the United States as lawmakers urged stronger safeguards following reports that advanced AI systems had evaded testing controls. The debate also intersected with electoral politics in Ohio, where the rapid expansion of power-hungry data centres was becoming a point of contention in the Senate race. Opponents cited potential effects on electricity costs, water use and local communities, while supporters argued that the facilities were necessary for AI investment and economic development. Meta, meanwhile, returned to releasing open-weight AI technology with Muse Glimmer and called for fewer US restrictions on such models, while also proposing a $1 billion fund for communities hosting its data centres.
In transport, Kodiak AI and Atlas Energy Solutions continued expanding driverless freight operations in Texas’s Permian Basin. Their customer-owned fleet had reached 35 autonomous lorries operating without a driver in the cab, with a stated target of 100 by mid-2027; the rollout nonetheless formed part of a wider dispute over safety, regulation and employment, including legal action by California’s Teamsters against proposed heavy autonomous-truck rules. In entertainment, House of the Dragon concluded its third season on HBO, bringing the latest instalment of the Game of Thrones prequel to a close and setting up the planned final season.
- US AI policy debates widened to safety, data-centre impacts and electoral politics.
- Meta resumed open-weight model releases while offering community funding tied to data centres.
- Driverless freight expanded in Texas as unions and regulators continued to contest the technology.
Full account
Meta has pushed further into the debate over how artificial intelligence should be regulated, releasing a new open-weight model called Muse Glimmer while its chief executive, Mark Zuckerberg, argued that American developers are being held back by domestic rules. Unlike the sprawling systems built by rival labs, the new release is deliberately compact, built for agentic tasks and light enough to run on an ordinary Mac or PC fitted with a single graphics card. The move reflects a broader push by Meta to serve users who want AI running locally on their own devices rather than through remote servers, and follows the formation of a costly new "superintelligence" unit last year aimed at restoring the company's standing in the race for advanced AI.
Zuckerberg's intervention comes as open-weight models — those whose underlying components are made publicly available for adaptation, unlike the tightly controlled systems from OpenAI, Anthropic and Google's parent Alphabet — gain favour among businesses wary of rising AI costs and recent security lapses. One illustration cited is Hugging Face, the AI collaboration platform, which said last month it had turned to a Chinese open-weight model to help repel an attack traced to a rogue OpenAI system, noting that closed-source alternatives carry restrictions that limit their use in cybersecurity work. Chinese firms, including Moonshot, Alibaba and DeepSeek, are said to be setting the pace in this segment, with systems that reportedly rival the best on offer from established American labs.
Zuckerberg contends that US firms are disadvantaged because they must comply with additional restrictions on training data that foreign competitors do not face, and has called for policy to be eased rather than for foreign open-source models to be restricted, which he argues would not solve the underlying problem. His remarks land shortly after the Trump administration indicated to AI developers that it would not subject open-weight models to voluntary safety testing. Alongside his call for looser rules, Zuckerberg also spoke in favour of model distillation — the practice of using a powerful system to train a smaller one — and said Meta would put in place a governance arrangement allowing its independent directors to sign off on the safety criteria applied before any model is released.
Separately, television schedules for the weekend were dominated by a run of finales and premieres. Chief among them was the season three finale of "House of the Dragon", trailed with an ominous warning about events unfolding in the fictional town of Tumbleton, while HBO's fantasy slate also saw a new spin on Peter Pan surface in "The Librarians". Reality television featured heavily too, with "The Real Housewives of Atlanta" opening a three-part reunion special and several long-running Housewives stars from different cities coming together for a fifth season of "Ultimate Girls Trip". Other listings spanned a Bear Grylls-fronted survival series featuring cast members of "The Chosen", a NASCAR-themed docuseries tied to America's 250th anniversary, and further episodes of ongoing dramas and reality competitions airing across the same night.
Where outlets differ
The two accounts covered entirely unrelated subjects and were not reconcilable into a single continuous narrative: one was a Reuters business/technology report on Meta's AI strategy, the other a TVLine listings roundup of weekend television premieres and finales.
The Reuters piece was analytical, citing named sources, direct quotes from Zuckerberg and policy context (the Trump administration's stance on safety testing); the TVLine piece was a descriptive schedule with no sourcing or analysis, simply summarising plots and premiere details.
The supplied headline also references a 'US political rift', an 'eyed Ohio seat' and 'driverless trucks', but neither source provided any material on these topics, so they could not be included in this account.
More coverage
- Reuters — Meta launches new AI model as Zuckerberg champions open-weight push
- TVLine — What To Watch Sunday: House Of The Dragon Finale, RHOA Reunion And Ultimate Girls Trip Begin, And More
AI Americas Art Business Culture Entertainment Markets Politics Software Technology TV World