Ten percent

I was 30,000 feet up, half-watching BBC News, when one headline made the rest of the flight feel very long: a senior AI researcher now puts the odds of AI ending humanity at more than 10 percent, within the next decade.

That is essentially what just happened in the AI industry. It deserves the full attention of ordinary Filipinos, not just Silicon Valley.

Evan Hubinger, a senior safety researcher at Anthropic, one of the world’s leading AI companies, said publicly that he believes there is more than a 10-percent chance AI kills all of humanity within the next decade.

He said it after a colleague, Jacob Coxon, resigned and accused Anthropic and OpenAI of racing toward autonomous, self improving AI despite fully understanding the danger. Hubinger didn’t deny it.

He confirmed it, and admitted Anthropic still has no working plan to keep a smarter than human AI system under control.
Today’s chatbots aren’t capable of this.

The danger these researchers describe is future facing: a system that improves its own intelligence, escapes oversight, resists being shut down, or replicates itself.

Not a robot uprising. A machine that outgrows us, quietly, faster than we can follow.

Ten percent sounds small. It isn’t. If someone told you there was a one in 10 chance your flight would crash, would you still board? No airline would let that risk stand. Yet that is the scale of concern now being voiced, not by outside critics, but by the people building the technology themselves.

This is no fringe opinion. Earlier this year, top AI leaders, including Anthropic’s own co-founders and OpenAI’s chief scientist, signed a statement demanding the power to deliberately slow frontier AI development. Their own warning: competition is pushing companies faster than safety research can follow.

The industry’s leadership is asking for brakes while its foot stays on the gas.

You might think this is a Silicon Valley story. It isn’t.AI already answers your loan inquiry, scans your face at the mall, and quietly runs inside hospitals, schools and banks across the country.

We are not spectators. We are users, customers and increasingly, dependents.

When the people building AI warn that development is outrunning our ability to control it, that is not an abstract worry. It is a preview of the world our children will inherit. A bank, fintech company, or hospital that adopts AI blindly, without asking who answers when something breaks, is gambling with public trust.

The House has already passed, on third and final reading, the National Cybersecurity and Critical Information Infrastructure Protection Act. It would create a National Cybersecurity Agency and require operators of critical systems, energy grids, banks, telecoms, and government networks to meet baseline security standards, undergo risk assessments, and disclose incidents when they happen. It now sits with the Senate.

This bill was written for today’s cyberattacks, not superintelligent AI. That is exactly why it matters as a first line of defense. Whatever shape the next AI-driven risk takes, the countries best positioned to respond will be the ones with clear rules, a coordinating authority, and mandatory incident reporting already built. The Philippines has none of that yet.

Every month this bill idles is another month our banks, hospitals, and utilities run on good intentions instead of an enforceable standard.

This bill won’t make the Philippines immune to the risks Hubinger and Coxon describe. But it’s a concrete step that doesn’t require solving the hardest problem in AI safety, only political will. Our legislators should pass it now.

I’ve spent much of my career pushing for financial inclusion through technology, because innovation done right expands human dignity, it doesn’t endanger it.

The lesson here isn’t “stop building AI.” It’s “build it the way we’d want a bridge built”: rigorous safety standards, independent oversight, real transparency, and the humility to slow down when the engineers themselves sound the alarm.

For Filipino policymakers, regulators, and business leaders, the takeaway is simple. Don’t wait for a crisis to ask hard questions about how AI is deployed in our financial institutions, our government services, our everyday platforms. Ask what safeguards exist. Ask who is accountable. Ask what happens the moment something fails. These aren’t anti-innovation questions.

They’re the questions that make innovation worth trusting.

The scientists who built this technology are telling us, in their own words, that they are not fully in control of where it’s headed. The least the rest of us can do is listen, and demand that safety finally catches up with ambition.

***

Lito Villanueva is the Philippines’ leading and award-winning thought leader in inclusive digital finance. As EVP and chief innovation and inclusion officer of RCBC, he has driven large-scale digital initiatives that advanced financial inclusion. He is the founding chairman of FinTech Alliance PH, representing 95 percent of the country’s digital retail financial transactions with an aggregate user base in excess of 100 million, and the co-founder and Global Ambassador of the Asia FinTech Alliance. Recognized as a People Asia Men Who Matter 2025, Digital Leader of the Year, Asia Trailblazer, and AGORA Awardee, he continues to shape the fintech landscape in the Philippines and beyond.