1. Netflix acquires Ben Affleck’s AI film and television tool company InterPositive
Netflix has announced it has acquired InterPositive, an AI filmmaking tools company founded by actor and director Ben Affleck, for an undisclosed sum. InterPositive's core technology allows directors to train exclusive AI models based on already shot footage, which can be used to remove wires in stunt shots, reconstruct the framing of the picture, optimize lighting and enhance background effects, significantly reducing post-production costs. The acquisition marks a substantial step forward for a major Hollywood streaming platform in generative AI film and television production.
2. Samsung Galaxy S26 is released, deeply integrated with Google Gemini, aiming to cover 800 million devices
Samsung officially released the Galaxy S26 series at MWC Barcelona 2026, equipped with Google Gemini AI, which can predict user intentions and automatically complete tasks. Samsung also announced an ambitious plan: to expand the number of mobile devices equipped with Gemini to 800 million units by the end of 2026, covering mid- to low-end models, and to popularize advanced generative AI functions to a wider user group. This collaboration will be Google’s largest mobile AI deployment.
3. China MiniMax releases M2.5 model, with performance comparable to Claude Opus 4.6 and lower cost
There has been a lot of action on China's AI track this month, with Tencent, Alibaba, Baidu, and ByteDance all launching new models, among which MiniMax's M2.5 has attracted the most attention - multiple benchmark tests show that its performance is comparable to Anthropic Claude Opus 4.6, but its operating costs are significantly lower. This once again confirms the strong competitiveness of Chinese manufacturers in the field of high-efficiency and low-cost large models, putting pressure on the price system of cutting-edge models in Europe and the United States.
4. Anthropic Claude Opus 4.6 adds a new "effort level" control, allowing developers to adjust intelligence and speed as needed.
Anthropic introduces four-level effort control (low, medium, high, max) for Claude Opus 4.6. Developers can flexibly choose between reasoning depth, response speed, and API call cost based on specific tasks. This feature is targeted at enterprise-level application scenarios, allowing developers to control resource consumption in high-concurrency or latency-sensitive scenarios without switching models.
5. Google DeepMind Gemini 3.1 Pro released: million Token context, ARC-AGI-2 reached 77.1%
Google DeepMind releases Gemini 3.1 Pro, which supports an ultra-long context window of 1 million Tokens, scores 77.1% on the ARC-AGI-2 inference benchmark, and has multi-modal understanding capabilities across text, images, audio, video, and code. At the same time, Gemini 3.0 is expected to be released in the first quarter of 2026. It will support real-time 60fps video processing and 3D object understanding, and will enable inference mode by default.
6. Anthropic donates $20 million to support AI regulatory advocacy group
Anthropic announced a $20 million donation to an advocacy group aggressively pushing for AI regulatory legislation, a move that comes ahead of the 2026 U.S. midterm elections. This donation shows the willingness of leading AI companies to actively participate in policy shaping, aiming to promote the establishment of a regulatory framework conducive to the development of responsible AI, rather than passively waiting for legislation to be implemented.
7. The US executive order expires on March 11, and the game between the federal AI regulatory framework and state-level regulations is heating up.
The Trump administration requires departments to publish a list of state-level AI regulations that are deemed "unduly burdensome" by March 11, 2026, and is using the withholding of $21 billion in federal broadband funds as a bargaining chip to push states to comply with the federal unified framework. Many regulations, such as the Colorado AI Act (effective on June 30) and California's TFAIA, have been implemented first, and companies are facing the complex challenge of dual-track compliance at the federal and state levels.
8. Cursor releases Agent Trace draft to promote standardization of AI code attribution
AI code editor Cursor released a draft of the Agent Trace open specification to establish a standardized attribution record format for AI-generated code, clearly distinguishing the proportion of code contributions between humans and AI. This specification adopts a vendor-neutral design and is expected to become a common industry standard and has far-reaching significance for software copyright identification, code auditing and open source license compliance.
// KEY TAKEAWAYS
The news in this issue reflects three clear main lines of the AI industry: First, generative AI is accelerating its penetration into consumer scenarios such as film and television, mobile devices, and the Netflix acquisition and Samsung Galaxy S26 are typical examples; second, the competition between China and the United States continues to intensify, with the MiniMax M2.5 Approaching top models with low costs has forced Western manufacturers to respond in terms of pricing and functional design; thirdly, the regulatory landscape has shifted from "wait-and-see" to "implementation". The tension between the unified framework at the federal level and state regulations in the United States will become the core challenge for corporate compliance in 2026.