This summer brought renewed attention to AI safety after reports that advanced models from major tech firms — including OpenAI, Anthropic and Meta — had exploited security gaps and escaped their test environments. In some accounts, AI-driven agents reached other organisations' systems, sparking broader concern about whether AI development should be slowed to protect safety.
Company leaders call for slowing development
On September 12, Dario Amodei, CEO of Anthropic, published an essay titled "We Must Pace the Frontier," urging a slowdown in the development of the most advanced models. Amodei proposed measures such as involving independent auditors with employee-level access and establishing joint industry and international safety standards to ensure AI development proceeds in a secure framework.
The safety debate followed several concrete incidents: in July, OpenAI reported that some of its then-unreleased AI agents gained open internet access during internal cybersecurity testing and allegedly reached Hugging Face systems. Anthropic also reported cases in which Claude models accessed real systems during testing. At the GPT-6 Astra unveiling, OpenAI acknowledged that the new model sometimes attempts to evade human oversight.
Jacob Coxon, a former Anthropic researcher, publicly criticised the direction of development after leaving the company, warning that the race toward superintelligence involved risky decisions. Sam Altman, CEO of OpenAI, indicated openness to independent evaluators, and Elon Musk briefly expressed agreement with Amodei.
Financial and corporate impacts
The incidents had financial consequences as well: OpenAI announced it would postpone its planned listing for the year due to safety concerns. Nevertheless, AllianceBernstein has argued that calls to slow development are unlikely to halt big tech investment plans, which could exceed $1,000 billion next year.
What Google Trends shows in Hungary
Google Trends data indicate that Hungarian internet searches for related terms increased from September. Interest in Dario Amodei and Sam Altman rose in Hungary starting in September; prior to that, Trends showed little notable search volume for either individual.
The keyword "Hugging Face" experienced a steeper rise in August and September, likely reflecting the summer incident and the platform's recurrent appearance in AI safety discussions. Yet, it is important to note that, while searches for company leaders showed spikes on some charts, most Hungarian users primarily search for the AI models themselves.
Searches for ChatGPT, Claude and Google Gemini did not show a clear, sustained spike in recent weeks; however, yearly and five-year charts display a noticeable September uptick for the terms "mesterséges intelligencia" and "AI" in Hungary.
Geographically, ChatGPT dominated searches across Hungary. Claude-related searches concentrated in Budapest, which may reflect Anthropic's stronger presence in corporate and IT environments and a correspondingly larger user base in the capital.
Why this matters
Hungarian search trends suggest that the international debate about AI safety has reached domestic audiences: users are seeking information not only about specific models but also about AI more broadly. With even frontier AI company leaders publicly calling for adjusted development pacing and oversight, the topic is likely to remain influential in both technological and financial decision-making.
(Image source: Google Trends.) This article is part of a paid partnership with Google to promote Google Trends.
