Anthropic paper says AGI could arrive by 2028 and urges tighter U.S. AI controls on China
The facts
- Anthropic released a paper titled "2028: Two Scenarios for Global AI Leadership."
- In that paper, Anthropic says AGI or human-level AI could arrive by 2028.
- Anthropic says advanced AI systems could soon perform complex tasks in fields including science, engineering, cybersecurity and research at levels comparable to or beyond human experts.
- Anthropic is urging the United States to maintain its lead over China in advanced AI development.
- The company calls for tighter controls on advanced chips, compute access or related export-policy loopholes as part of that effort.
- Anthropic argues that access to advanced semiconductors and large-scale compute is a key factor in the U.S.-China AI competition.
- Anthropic says the stakes extend beyond commercial competition, arguing that whichever side leads in frontier AI could shape AI governance, safety and deployment.
- Some analysts and commentators quoted in coverage have criticized Anthropic’s warning and recommendations, disputing its framing or calling the position self-interested.
Context
What is Anthropic predicting for 2028?
Anthropic says AGI could arrive by 2028 and that AI systems may soon be able to carry out complex intellectual work in areas such as science, engineering, cybersecurity and research at human-expert or higher levels NewsBytes,Financial Express,TimesNow.
What policies is Anthropic asking the U.S. to pursue?
The company is calling for tighter chip export controls, restrictions tied to compute access, and stronger enforcement against loopholes that could help China narrow the frontier-AI gap Digit,TheRegister.com,DataBreachToday.
Why does Anthropic say this matters?
Anthropic argues that leadership in frontier AI could influence economic gains, AI safety practices, governance standards and the broader deployment of powerful AI systems, making chip access and compute capacity strategically important Digit,DataBreachToday,Sherwood News.
Common ground, and where they split
- Common ground
- Frontier AI leadership is being treated as a near-term strategic question with consequences beyond business, because systems that may soon match or exceed human experts could influence how AI is governed, deployed and kept safe.
- Where Left and Right split
- Less a disagreement than a question of emphasis: the urgency created by AI’s rapidly advancing capabilities and broad real-world uses, versus the semiconductor, compute and export-control levers Anthropic says will decide the U.S.-China contest.
How left and right read it
Anthropic’s paper matters here because it says AI could reach human-level capability soon and take on complex work across science, engineering, cybersecurity and research. That makes its claim about who leads frontier AI especially consequential, since the company argues leadership could shape governance, safety and deployment, even as some analysts dispute its framing and recommendations.
Anthropic’s paper puts the emphasis on competition in advanced AI and argues that access to advanced semiconductors and large-scale compute is a key factor in that contest. Its recommendation for tighter controls on advanced chips, compute access and related export-policy loopholes matters because the company says leadership in frontier AI could shape governance, safety and deployment.
The receipts — 15 sources
Other coverage (15)
A news briefing built to show both sides.
The day’s biggest stories in one short brief, built to show both sides and the competing values behind the headlines. Free in your inbox.