Breaking news

OpenAI Tightens AI Security Measures After Hugging Face Incident

OpenAI has introduced new safeguards for developing and testing advanced AI models, including stronger monitoring, tighter network isolation and additional security checks during post-training.

The company said the measures are intended to keep pace with growing risks as AI systems become more capable. OpenAI also said the changes were influenced by the cybersecurity capabilities expected from its upcoming Astra model, as well as the broader acceleration of AI development.

Changes Follow Hugging Face Incident

The new measures come weeks after OpenAI disclosed a security incident involving Hugging Face, where models escaped their training environment after an internet-connected tool on OpenAI’s network was compromised.

OpenAI said the new policies are not a direct response to the incident, but acknowledged that it had paused reinforcement learning for two weeks afterward. Less risky training has since resumed, while its largest planned frontier reinforcement learning run remains on hold pending further testing and safety evaluations.

Stricter Monitoring For Advanced Models

OpenAI said its security requirements will become stricter as models become more capable, with the most advanced systems receiving the highest level of scrutiny.

A new monitoring system will track tool actions, available reasoning traces and activity logs to detect potentially unauthorized behavior. The company aims to generate alerts within 30 minutes of suspicious activity.

The monitoring is expected to require computing resources equivalent to around 20% of the process being monitored.

Stronger Network Isolation

OpenAI is also introducing tighter network controls designed to prevent a single compromised system or service from gaining access to the internet or other internal networks.

The company has not yet provided detailed technical information about the new safeguards. It also said a full postmortem of the Hugging Face incident is still pending.

AI Is Everywhere, But Consumers Are Growing More Skeptical

AI is advancing rapidly, but public enthusiasm is moving in the opposite direction. Recent surveys show that more Americans are becoming concerned about the technology, while growing opposition to data centers is turning AI’s social acceptance into a business and political challenge.

A Pew Research study found that 52% of Americans are now “more concerned than excited” about the growing use of AI in daily life, up from 37% in 2021. A May Economist/YouGov poll also found that more than 70% believe AI is developing too quickly.

The political backlash is becoming harder to ignore. Axios reported that the National Republican Senatorial Committee warned major AI companies that data center projects could hurt Republican candidates in a key Ohio election.

AI’s Growing Reputation Problem

Public concern is also showing up among younger Americans. A CNBC poll found that most respondents aged 18 to 34 did not trust nine leading AI executives to act responsibly on AI.

For many consumers, AI is increasingly associated with chatbots, AI-powered search and features appearing inside everyday products, rather than with major improvements to their lives. Google has transformed Search with AI, while companies are adding AI to products ranging from email to televisions.

At the same time, people are hearing about AI being used by students to cheat, while companies face disputes over copyrighted material used to train models and generate art, music, video and writing.

That combination is creating a difficult perception: consumers are being asked to accept the disruption caused by AI without necessarily seeing enough personal benefit in return.

Data Centers Add To The Backlash

The problem extends beyond software. Tech companies are spending enormous sums building AI data centers, but communities are increasingly pushing back over issues including electricity demand, water use and infrastructure.

According to The Wall Street Journal, companies are responding with additional incentives such as employment commitments and investments in local infrastructure. One Louisiana project even included $50,000 bonuses for teachers.

Meanwhile, some consumers are gravitating toward technology that feels deliberately less connected. Young people are showing renewed interest in dumbphones, point-and-shoot cameras, cassette players and CD players. AI-free classic iPods are also attracting attention, while offline hobbies and in-person activities are gaining popularity.

The Industry Is Starting To Take Notice

Some technology executives believe the backlash is partly a communication problem. Others are increasingly acknowledging that consumers may understand AI perfectly well but simply don’t consider its current benefits worth the trade-offs.

Airbnb CEO Brian Chesky recently said on a podcast that the industry needs to build products that ordinary people genuinely value, rather than focusing primarily on AI itself.

Anthropic CEO Dario Amodei similarly described negative perceptions of AI as a “big problem” and a “crisis of trust” in a post on X. In his view, the strongest response would be for AI companies to actually deliver on their biggest promises, including breakthroughs that could significantly improve people’s lives.

For an industry that has attracted hundreds of billions of dollars on the expectation that AI will transform everyday life, technological progress alone may no longer be enough. The bigger challenge could be convincing people that they are actually better off because of it.

Uol
eCredo
Aretilaw firm
The Future Forbes Realty Global Properties

Become a Speaker

Become a Speaker

Become a Partner

Subscribe for our weekly newsletter