Some of the companies competing to build the most advanced AI models are now calling for development to slow down. Their intervention is drawing scrutiny of their safety commitments and commercial interests.
On Saturday 12 September, Anthropic CEO Dario Amodei published “We Must Pace the Frontier”, arguing that the development of increasingly powerful ‘frontier’ AI models needs to be paced so that safeguards have time to catch up.
Amodei’s proposals include three key measures:
Anthropic has committed to introducing evaluators and OpenAI CEO Sam Altman pledged to follow, while Elon Musk and Google DeepMind’s Demis Hassabis expressed support. For now, there is, however, no agreed industry-wide slowdown.
Scepticism has been a recurring theme in European coverage, with scrutiny focusing on AI companies’ motivations and whether they can credibly police themselves. The warnings are receiving attention and are taken seriously, but so are the interests behind the companies making them.
In the UK, the week began with extensive, predominantly news-led coverage following Amodei’s post, with live blogs by national and broadcast outlets such as Independent, Sky News and BBC, tracking corporate reactions and political developments. However, alongside this, The Guardian questions whether commitments will lead to meaningful change, highlighting concerns that industry-designed rules could in fact favour the established AI companies.
The question of who should provide oversight is also gaining attention. In a parallel development on 14 September, the UK’s Joint Committee on Human Rights, which brings together members of both houses of Parliament, called for a new AI Bill and an independent oversight body.
Similar concerns about accountability run through German coverage. National business newspaper Handelsblatt examines Amodei’s proposals and the incidents behind his warning, while DIE ZEIT questions evaluators’ independence and whether companies are seeking to shape future regulations on their own terms. Safety warnings are being taken seriously, but their commercial context is also being heavily scrutinized.
Experts quoted in the Netherlands are more openly doubtful about corporate motives. In Trouw, an AI and society professor argues that the warnings serve to promote companies’ capabilities to attract more investments. Similarly, an AI security researcher quoted by de Volkskrant similarly interprets the intervention as an effort to position Anthropic as an ethical AI company ahead of a possible stock market listing.
In France, coverage extends these questions to the practical feasibility of oversight. Tech publication 01.Net questions the legitimacy of industry-led supervision, while Les Echos examines the challenges posed by commercial and geopolitical competition. La Tribune offers a counterpoint, exploring the argument that Europe must accelerate AI adoption to avoid falling behind.
This tension between caution and competitiveness is also shaped by the current geopolitical backdrop. US President Donald Trump’s dismissal of the warnings, while prioritizing America’s lead over China, illustrates why agreement among AI executives may not translate easily into a coordinated slowdown.
Terms such as “fully autonomous” or “self-improving” could invite questions about what a system is allowed to do, when a person must intervene and how it can be stopped.
For upcoming product announcements, this means supporting claims about speed and capability with evidence of safeguards. Journalists may increasingly expect details of how AI agents are evaluated, the limits placed on their use, and the outcomes achieved for customers.
This scrutiny could also create opportunities for security, engineering and governance spokespeople to explain incident disclosure, human oversight and accountability, through interviews, podcasts and keynote events. Practical media and speaker training should prepare them to answer challenging questions clearly, across different channels.
Policy communications will remain important as company commitments to AI safety and government responses develop. This will encourage greater focus on AI regulation, and how this will impact businesses in the coming years.
Pauline Delorme is Director at Tyto, based in the UK. In this role, Pauline focuses in developing and maintaining a network of key journalists and influencers across a variety of sectors.