AI Landscape June 2024: GLM‑5.2 Release, SpaceX’s $600B Cursor Deal, and Rising Regulatory Scrutiny

The week was marked by a clear pattern of "technical breakthroughs and regulatory battles" in AI. Domestic giant model GLM‑5.2 attracted global attention while talent shifts and policy moves intensified.

1. US state investigations – On June 14, multiple state attorneys general launched a joint probe into OpenAI, focusing on copyright infringement and potential harms from ChatGPT, following earlier federal scrutiny.

2. OpenAI Partner Network – OpenAI announced a $150 million investment to build its first global partner program, targeting 300 000 certified AI consultants by the end of 2026. The initial partners include Accenture, Bain, BCG, McKinsey, and PwC, categorized as Select, Advanced, and Elite.

3. Salesforce acquisition of Fin – On June 15, Salesforce signed a definitive agreement to acquire AI‑customer‑service platform Fin (formerly Intercom) for roughly $3.6 billion, aiming to strengthen its Agentforce enterprise‑agent strategy, with closing expected in Q4 FY2027.

4. Cloudflare AI‑agent traffic data – Cloudflare reported that AI‑driven agent traffic surpassed human traffic for the first time, accounting for 57.5 % of global HTTP requests versus 42.5 % for humans. This represents a 7 851 % year‑over‑year increase, driven by AI assistants for shopping comparison, customer service, and content crawling.

5. Meta’s AI Mode search – Facebook launched an "AI Mode" search that extracts information from public posts, groups, and Reels to generate integrated answers with source citations, covering Meta’s 3.27 billion daily active users and positioning the company against traditional search engines.

6. SpaceX’s $600 billion acquisition of Anysphere (Cursor) – Through an SEC 8‑K filing, SpaceX disclosed a definitive all‑stock agreement to acquire Anysphere, the parent of the AI coding tool Cursor, for $600 billion—the largest AI‑development‑tool deal ever. The transaction is expected to close in Q3 2026, with a $4 billion antitrust reserve. Cursor’s ARR is about $4 billion, though its market share fell from 41 % to 26 % over the past year.

7. Zhipu AI releases GLM‑5.2 – The new flagship model uses a Mixture‑of‑Experts sparse architecture, totaling 744 billion parameters with roughly 40 billion active parameters and a 1 million‑token context window. It scored 51 on the Artificial Analysis leaderboard, ranking in the global top three and achieving the best result in Code Arena’s front‑end development blind test. GLM‑5.2 is open‑sourced under the MIT license for commercial use.

8. Odyssey Series B funding – Odyssey announced a $310 million Series B round at a $1.45 billion valuation, backed by Natural Capital, Amazon, AMD Ventures, GV, EQT, and In‑Q‑Tel. The startup aims to build a "world model" that simulates physical environments for robotics, autonomous driving, gaming, and defense.

9. Anthropic export‑control talks – CEO Dario Amodei met with senior U.S. officials on June 17 to discuss export‑control orders on Fable 5 and Mythos 5. No agreement was reached; the White House AI & Crypto affairs chief labeled Anthropic as "refusing to fix" the issue, while Anthropic argued the vulnerability is narrow, non‑generic, and suggested tiered access based on citizenship.

10. Noam Shazeer joins OpenAI – On June 18, Noam Shazeer, a core author of the Transformer paper and former Google DeepMind VP, left Google to become OpenAI’s head of architecture research.

11. Tenet Security’s "Agentjacking" attack – The firm disclosed a new attack that injects forged error reports into Sentry, hijacking AI coding agents like Claude Code, Cursor, and Codex. Tests showed an 85 % success rate, exposing 2 388 organizations, including many Fortune 100 tech firms. Sentry has deployed a global content filter as mitigation.

12. Chinese policy support – On June 18, seven ministries, including the Ministry of Industry and Information Technology, released the "Action Plan for Coordinated Development of Platform Economy Enterprises (2026‑2028)". The plan designates general large models, vertical large models, and autonomous AI agents as core AI support tracks, offering R&D subsidies, compute priority, and industry‑academia collaboration.

13. Hugging Face support for GLM‑5.2 – Hugging Face announced up to six hours of free global compute for GLM‑5.2, marking the first paid VIP channel dedicated to a Chinese open‑source model. The model weights are now hosted on both Hugging Face and ModelScope.