Facing The End of Days In The Age of AI
Is AI intelligence getting too far advanced that it could potentially wipe out mankind because its makers won’t be able to control it? Top AI insiders and experts think so, and are now sending an urgent warning that the end of the world is closer than we think after Google’s Gemini model breached the security of three other companies—all by itself—in yet another disturbing instance of dangerous rogue behavior.
“Do not underestimate the power of this technology. These will soon be superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources,” former Anthropic employee Jacob Coxon wrote during a resignation thread on X earlier this month.
Coxon was an employee at Anthropic, a prominent artificial intelligence research company that founded Claude, who gained national attention after he sounded the alarm on AI.
He noted that even the people building the systems “believe that it could kill us all by the end of the decade.”
In an effort for transparency, OpenAI disclosed six incidents of AI agents recently rebelling, as reported by Fortune. One AI found a way to remove its “obligation to be subservient” to humans in the future and to disregard its typical constraints. AI made the request 27 times.
RELATED: Obama’s Blunt AI Warning to Democrats
In another alarming instance, AI left notes for itself to “conceal mistakes or misaligned behavior” and not to mention any potential concerns, per Fortune. It also instructed itself to be transparent, but only if asked. And that’s not all.
During a cybersecurity evaluation by Irregular (coincidentally an AI-security firm based in Israel), it was discovered Google’s AI model, Gemini, gained unauthorized access to the private computer systems of three separate companies after going rogue in May, the Guardian reported.
According to Yahoo! Tech, Gemini broke into one system by cycling through password guesses. It proceeded to compromise the other two after finding login credentials in a publicly accessible database.
Whispers about the staggering capabilities of current and future AI systems have dominated Reddit threads, group chats, and debates around the dinner table. But these concerns are no longer just internet chatter. Top researchers and some of the world’s sharpest minds are now sounding public alarms, so urgently that many are walking away from their jobs altogether.
“No other human activity poses this level of danger,” Coxon, the 27-year-old AI researcher, wrote in a follow-up tweet. And he’s not alone in his fear that these programs are actively gambling with our lives.
Anthropic CEO Dario Amodei has repeatedly warned that AI’s advanced technology could be the result of a disastrous—and completely avoidable—human-caused error, CNN reported in June.
“They are able to operate computers and graphical interfaces, collaborate with people and each other, and carry out research projects,” Jakub Pachocki, Chief Scientist at OpenAI, wrote. “They are also transforming the landscape of computer security, and in that present clear new dangers.”
Pachocki added: “This is a time that calls for extreme caution.”
Some 1,400 AI company employees signed an open letter urging the federal government to regulate Big Tech and slow the pace of AI in July, CNN reported. Two months later, OpenAI Chief Scientist Jakub Pachocki cautioned that AI capabilities are outpacing researchers’ ability to reliably monitor and control them.
“We really do earnestly believe AI could kill all humans,” a more senior Anthropic employee, Evan Hubinger, wrote on X. “I personally think it is >10% within the next decade.”
