The morning briefing.
While you slept: Anthropic warns of a critical AI race with China, OpenAI faces a supply chain hack, and MediaTek pushes smartphone AI agents.
Geopolitical AI. Anthropic has issued a stark warning to Washington, emphasizing that the current period is a critical "now-or-never" moment for the US to secure its compute lead over China in the AI race. This geopolitical competition is further complicated by reports of overseas actors using AI-generated videos to spread narratives of UK decline. The UK is also exploring options for sovereign LLM inference to bolster national AI capabilities.
AI Security. The vulnerability of AI systems and their supply chains is a growing concern, as security researchers, reportedly aided by Anthropic's Mythos, claim to have breached macOS. Separately, OpenAI experienced an internal security incident where employee devices were compromised through a poisoned npm package, leading to the theft of limited credential material. Persistent threats from groups like Chinese hackers MustangPanda continue to target victims across Asia.
On-Device AI. The push for AI at the edge is accelerating, with MediaTek positioning its Dimensity platform as crucial for enabling smartphone AI agents, citing a 7x surge in daily autonomous AI tasks. Apple is reportedly testing Intel's 18A-P process for its future iPhone and Mac chips, indicating a focus on advanced manufacturing for on-device capabilities. Meanwhile, SMIC reports that the AI boom is driving significant demand for power management and data transmission chips, and OpenAI has made its Codex coding assistant available on iOS and Android.
AI Investment. Major players continue to pour resources into AI, with Tencent reporting a 9% revenue growth in Q1 2026 and signaling significant AI capital expenditure expansion throughout the year. Beijing startup DeepCybo has raised hundreds of millions of RMB for embodied AGI, validating a trend seen with Tesla and NVIDIA. In the UK, EdTech firm Multiverse secured €60 million to expand AI and tech adoption training, while the UK's CMA has opened an antitrust inquiry into Microsoft's dominance in business software.
Anthropic warns Washington on AI competition with China
Anthropic has published a policy paper urging Washington to act decisively to maintain its lead over China in AI compute, framing the current period as a critical juncture. The paper outlines scenarios where the US either secures its position or authoritarian regimes dictate the future of AI by 2028.
DeepCybo raises hundreds of millions for embodied AGI with first-person data
Beijing AI startup DeepCybo, founded by Chen Kai, has secured hundreds of millions of RMB to pursue embodied AGI using first-person human video data. This approach has recently been validated by major players like Tesla, NVIDIA, and Figure AI adopting similar strategies.
Cerebras stock falls after blockbuster IPO debut
Cerebras Systems' shares experienced a decline in early trading on Friday, following its strong stock market debut on Thursday. The company, known for its AI chip technology, is navigating post-IPO market dynamics.
Security researchers claim macOS breach with Anthropic's Mythos AI
Security researchers, reportedly assisted by Anthropic's Mythos AI, claim to have successfully breached macOS, prompting Apple to take the claims seriously. This incident highlights the evolving role of AI in cybersecurity exploits.
AI boom drives demand for power management chips, says SMIC
SMIC co-CEO Zhao Haijun reported that the AI boom is significantly increasing demand for power management and data transmission chips. This trend is tightening memory supply chains and boosting orders for the company's automotive-grade analog BCD platform.
MediaTek Dimensity platform powers smartphone AI agents
MediaTek's latest Dimensity developer conference highlighted the platform's role in enabling smartphone AI agents, with daily autonomous AI task volume surging sevenfold year-over-year to 870 million in 2026. The company is positioning its chips as key for the next generation of mobile AI.
Chinese hackers MustangPanda return to target victims across Asia
Researchers have detected an updated version of the FDMTP backdoor being deployed by Chinese hacking group MustangPanda through DLL sideloading. The group continues to target victims across Asia, demonstrating persistence in its cyber operations.
OpenAI employee devices compromised in npm supply chain attack
OpenAI experienced a security incident where malware hidden in poisoned npm packages reached two employee machines, resulting in the theft of a limited amount of internal credential material. The company is addressing the supply chain vulnerability.
Apple reportedly testing Intel's 18A-P process for iPhone and Mac chips
Apple is reportedly testing Intel's 18A-P process technology for manufacturing future iPhone and Mac chips. The initial production run is expected to focus on older Apple Silicon systems.
Overseas fakers use AI videos to push UK decline narrative
The BBC has uncovered evidence of overseas actors using AI-generated videos to promote a narrative of decline in the UK. This highlights the growing challenge of AI-powered misinformation campaigns.
Britain's latest civil servant is a chatbot trained on GOV.UK
Whitehall has introduced a new AI assistant, trained on GOV.UK data, to help citizens navigate public services more efficiently. The initiative aims to streamline interactions with government resources.
UK opens antitrust enquiry into Microsoft's business software dominance
The UK's Competition and Markets Authority (CMA) has launched an antitrust inquiry into Microsoft's dominance in the business software market. The investigation will examine the company's practices and their impact on competition.
OpenAI makes its AI coding assistant Codex available on iOS and Android
OpenAI has expanded the availability of its AI coding assistant, Codex, to the ChatGPT app on both iOS and Android platforms. This move brings advanced coding capabilities to mobile developers.
Tencent Q1 2026 revenue up 9% as AI investment surges
Tencent reported a 9% year-over-year revenue growth to RMB 196.46 billion in Q1 2026, with management indicating a significant expansion of AI capital expenditure throughout the year. The company is heavily investing in AI integration across its ecosystem.