എഐ സാങ്കേതികവിദ്യയുടെ സുരക്ഷയ്ക്കായി ഓപ്പൺ എഐ, ഗൂഗിൾ, മെറ്റ തുടങ്ങിയ വമ്പൻ കമ്പനികൾ വൈറ്റ് ഹൗസുമായി കരാറിലേർപ്പെട്ടു. എന്നാൽ, ഇതൊരു നിർബന്ധിത നിയമമല്ലെന്നും കമ്പനികളുടെ സ്വയം നിയന്ത്രണം മാത്രമാണെന്നും വിമർശകർ ചൂണ്ടിക്കാട്ടുന്നു. ലാഭേച്ഛയോടെ പ്രവർത്തിക്കുന്ന കമ്പനികൾക്ക് ഇത്തരം സുരക്ഷാ മാനദണ്ഡങ്ങൾ പാലിക്കാൻ കഴിയുമോ എന്ന ആശങ്കയാണ് ഇപ്പോൾ ഉയരുന്നത്.
Here's Tech-AI-ila, your weekly dose of what's changing in tech, AI and startups.
"Your scientists were so preoccupied with whether they could, they did not stop to think if they should."
That line from Jurassic Park was written for a film about dinosaurs. But cut to 2026, and it feels like it could have been written for Silicon Valley. Because this week, some of the companies racing to make AI more powerful also agreed that perhaps somebody should put a few guardrails around it.
Six of the biggest names in AI - OpenAI, Anthropic, Google, Meta, xAI and Nvidia - signed the White House Accord on Super Intelligence, a voluntary agreement aimed at strengthening safety around increasingly powerful AI systems.
And that leaves us with a rather awkward question: Can the companies building the most powerful AI systems also be trusted to police them?
The accord sounds reassuring enough. It calls for internal safety controls, monitoring, testing, independent assessments and oversight by an independent committee of each company's board. On paper, quite a lot of grown-ups are watching the children.
But then comes the small print.
The agreement is voluntary. There is no new regulator, and no legally enforceable punishment for a company that fails to live up to its promises. And that is where things get interesting. A promise is one thing. What happens when someone breaks it is another.
Connor Martin, a former deputy director at the US Treasury's Committee on Foreign Investment, has raised precisely this concern: voluntary commitments may struggle to change the commercial incentives pushing companies to build increasingly powerful AI as quickly as possible.
David Sacks, a Silicon Valley venture capitalist turned Trump's tech Czar, sees it differently. He argues that the accord has more teeth than the word "voluntary" suggests, pointing to independent audits, board oversight and existing corporate, FTC and securities laws that could still hold companies accountable.
Which brings us to the Delhi version of the debate: "Do kachori, ek samosa - in companies ka kya bharosa?"
Jokes aside, that's really the question. If the companies building frontier AI are also responsible for keeping it safe, who watches the watchdog?
Interestingly, the companies themselves don't necessarily disagree that stronger safeguards are needed. OpenAI has previously supported mandatory, capability-based national AI safety requirements, while Anthropic's Dario Amodei has called for greater oversight and independent testing of powerful AI systems.
So the argument isn't really about whether AI needs guardrails. It's about who builds them, who checks them and what happens when someone ignores them.
And there's another part of this story that businesses should probably pay attention to. The responsibility may not stop with the companies building AI.
Gartner analysts John Lovelock and Mark McDonald have pointed out that the accord also puts greater responsibility on businesses deploying these systems.
That matters because AI is rapidly moving from something you chat with to something companies actually use - writing code, dealing with customers, making decisions and increasingly carrying out tasks on its own.
So the question for a company may soon change from "Which AI model are we using?" to "What have we put around it to make sure it doesn't do something stupid?" And that is a much more uncomfortable question.
It matters in India too. India may not be at the forefront of building the world's most powerful AI models, but Indian companies are already deploying AI across everything from customer service to software and finance. That means India has a pretty obvious interest in how the rules for powerful AI are written.
If Washington relies mainly on voluntary commitments while other countries build formal regulatory systems, companies operating across borders could soon find themselves dealing with a patchwork of different rules on safety, transparency and accountability. India has a choice: simply adopt whatever rules Washington, Brussels or Beijing eventually come up with, or have a meaningful voice in shaping them.
Meanwhile, Washington has renamed AI
Because apparently one AI debate wasn't enough.
A September 29 executive order directed US federal agencies to use "Super Intelligence", or "SI", instead of "Artificial Intelligence", or "AI", in official communications and other non-statutory documents. Apparently even the name needed an upgrade.
But Shakespeare had a point when he asked: "What's in a name?"
The bigger question isn't whether we call it AI or Super Intelligence. It's whether the people building increasingly powerful machines should also be the ones deciding how powerful is too powerful.
And that brings us neatly back to Jurassic Park. The difficult part was never proving that we could build something extraordinary. It was figuring out what to do once we had.