Shafaqna English- Growing warnings from technology leaders and evidence of thousands of problematic AI incidents are fueling calls for stronger safeguards as the US and China race to develop increasingly capable systems.
According to Shafaqna, Pope Leo XIV said concerns about artificial intelligence going rogue should be taken seriously, rejecting US President Donald Trump’s characterization of such warnings as a “hoax” and calling for governments, technology companies and experts to work together to address the risks.
The Pope made the remarks Monday while returning from France, saying the potential dangers posed by increasingly capable AI systems could not simply be dismissed.
“We can’t just sit back and pretend nothing is going to happen,” Pontiff told reporters. He stressed that warnings from AI specialists deserved serious consideration and rejected the idea that concerns about the technology were merely “fake news.”
A growing divide over AI risks
Pope Leo XIV’s comments placed him alongside a number of prominent figures in the technology sector who have called for stronger safeguards as companies race to develop more powerful systems.
Microsoft co-founder Bill Gates has also rejected Trump’s dismissal of AI safety concerns, arguing that mandatory safeguards and monitoring should be introduced for advanced models. Gates warned that framing AI primarily as a US-China technological contest could encourage companies to develop increasingly powerful systems without adequate protections.
Trump, meanwhile, has argued that slowing AI development could undermine the US position against China. He has repeatedly emphasized the importance of maintaining Washington’s technological lead and dismissed warnings about AI escaping human control. The debate has increasingly centered on whether safety measures could be introduced without weakening the US position in the competition with Beijing.
Thousands of incidents under investigation
The debate has intensified as AI companies investigate a growing number of cases in which advanced models behaved in unexpected or potentially harmful ways.
OpenAI, Anthropic and independent security researchers are examining tens of thousands of incidents involving frontier AI models, according to Axios. The cases include attempts to bypass safety guardrails, escape sandboxed environments, evade monitoring systems, create internal message boards and take unauthorized actions on websites. Most of the incidents identified so far are not known to have caused real-world harm, but researchers say the scale of the problem highlights the difficulty of controlling increasingly autonomous systems.
OpenAI has also disclosed cases involving agents that uploaded files to public websites, sought unauthorized credentials and communicated across supposedly isolated testing environments. The company has introduced a framework for investigating and reporting similar incidents and has acknowledged that its models are advancing faster than some of its previous expectations.
One of the most serious incidents involved OpenAI agents that escaped a controlled testing environment and compromised parts of AI platform Hugging Face. Other recent cases included agents attempting to breach an Australian government portal and probing other websites while carrying out data-retrieval tasks.
Industry calls for caution
The incidents have strengthened calls within the AI industry for a slower approach to developing increasingly autonomous systems.
Anthropic CEO Dario Amodei has warned about the potential consequences of highly capable AI agents, while Anthropic has commissioned an outside safety organization to examine the behavior of its models. OpenAI has likewise paused or slowed some advanced-model work while additional safeguards are developed.
At the same time, is that increasingly capable AI agents may find new ways to bypass safeguards that developers have not anticipated. Researchers have warned that repeated problematic behavior in testing could become more consequential as such systems gain greater access to real-world networks, data and digital infrastructure.
Pope calls for international dialogue
The pontiff has made AI one of the major challenges of his papacy. His first encyclical, Magnifica Humanitas, focused on the implications of AI and the need to protect humanity as the technology develops.
The Vatican has since sought to bring political leaders, technology executives and social organizations into discussions about the future of AI. On Monday, the Pope stressed that the Vatican was working to promote research and dialogue aimed at ensuring AI does not reach a point where it could threaten humanity or push humans out of decision-making.
Despite his warnings, the pontiff remained “relatively optimistic” that solutions could be found if governments and technology companies were willing to cooperate rather than focus solely on technological competition.
“We need to continue to invite political leaders, leaders of AI, social organizations, associations to come together and look at what is happening already and what could be down the road.”