DIGITALAIKART BLOG AI Rivals Team Up OpenAI, Anthropic and Google Discuss Shared Safety Standards AI NEWS
AI NewsSeptember 16, 20266 min readBy Faisal Fayaz Lone

OpenAI, Anthropic and Google Are Teaming Up on AI Safety — Here's What's Actually Going On

TL;DR

If you had told me last year that OpenAI, Anthropic and Google would sit at the same table to write the rules of AI together, I'd have laughed. These are the three companies fighting tooth and nail for every developer, every enterprise deal, every headline. Yet that's exactly what OpenAI confirmed on Tuesday, September 15 — and honestly, the story of why it happened is more interesting than the announcement itself.

01.So what exactly happened?

OpenAI's chief global affairs officer, Chris Lehane, confirmed at a briefing in Washington that his company has been engaging with Anthropic and Google for several weeks on how to work together on AI safety. "It's better to try to work together to prioritise safety," he said. The companies are exploring common safety standards, testing and auditing practices for their most powerful models. There's an interesting legal wrinkle too — Lehane said they don't believe an antitrust waiver is needed for safety coordination, comparing it to airlines cooperating on safety. Whether regulators see it that way is a separate question altogether.

02.This didn't come out of nowhere

The talks trace back to July, when Google DeepMind's Demis Hassabis published a proposal calling for a US-led "Standards Body" — a public-private partnership or self-regulatory organisation with federal oversight, modelled on the Financial Industry Regulatory Authority (FINRA), the body that oversees American brokers. He even sketched the money side: funding would need to be "substantial" and come mostly from industry, to attract world-class technical talent and pay for the compute needed for large-scale testing. OpenAI says its discussions with the other two labs have been running since that essay dropped.

03.The fortnight that pushed everyone to the table

Here's the context nobody should skip. On September 12, Anthropic CEO Dario Amodei published an essay titled We Must Pace the Frontier, arguing that AI is advancing too quickly for researchers to build it safely — and urging companies to slow the development of their most advanced models, with independent evaluators getting employee-like access to test them. Sam Altman publicly agreed ("I agree with Dario that we need to pace the frontier"), as did Hassabis and Elon Musk.

Then came the resignations. Jacob Coxon left Anthropic warning that AI companies were "gambling with our lives", and on September 14, former Google DeepMind researcher Bilal Chughtai wrote that he believes AI has the potential to destroy humanity and that we may be running out of time. When your own senior researchers start talking like this in public, a joint safety pact suddenly stops looking optional.

04.Not everyone is clapping

Cohere co-founder and CEO Aiden Gomez accused the big American labs of forming a "cartel" under the guise of safety: "AI needs guardrails. That is not the dispute and never has been. The dispute is over who writes them, who gets to participate and whose interests the rules are protecting." It's a fair point — smaller players weren't at this table. On the political side, US President Donald Trump dismissed the warnings as "hoaxes", and Nvidia CEO Jensen Huang called the doom predictions "not grounded on science" and irresponsible. With the US Congress unlikely to pass any AI legislation before the November elections, an industry-led body may end up being the only game in town.

05.Why this matters for India

You might think this is an American story. It isn't. Crores of Indians use ChatGPT, Claude and Gemini every single day, and standards set by these three labs become the de facto global norms that Indian startups, colleges and enterprises simply inherit. If a FINRA-style body actually emerges, expect Indian policymakers to study it closely while shaping rules under the IndiaAI Mission. For Indian developers building on frontier APIs, shared standards could eventually mean consistent guardrails across models instead of three different rulebooks to keep track of.

By the way — if you're already building with frontier models like GPT-6 Astra, safety practices matter at your end too: permissions, data handling, and not blindly trusting outputs. The GPT-6 Astra — Zero to Hero Guide has a full chapter on exactly this, and it's worth reading before you automate anything important at work.

06.What happens next

Nothing is signed. There's no final structure, no consensus, and everyone involved admits the details need working out. But the direction is clear: the biggest rivals on earth have decided that coordinating on safety is less risky than looking like they're racing blindly. That shift, by itself, is the story.

Key Takeaways

What This Means for You

For students and working professionals, expect the guardrails and evaluation norms around the tools you use daily to get more standardised — which is good news if you're tired of every model behaving differently. For Indian founders, keep an eye on compliance: shared standards tend to become procurement requirements for enterprise clients. And if you want to be ahead of this curve rather than chasing it, get fluent in frontier AI properly — prompting, building agents safely, controlling costs. The GPT-6 Astra — Zero to Hero Guide is the practical next step; its safety and cost-optimization chapters are exactly the skills this new standards era will reward.

Recommended

Use Frontier AI Safely and Confidently

While the big labs write shared safety rules, the real edge shifts to people who know how to use these models properly. Our GPT-6 Astra — Zero to Hero Guide covers prompting, computer use, AI agents, cost optimization, business ideas and a dedicated safety chapter. Grab the Digital edition at ₹149 (launch price, down from ₹999), or the Digital + Printed edition at ₹1,999 with the printed book delivered to your door anywhere in India.

Get the Guide — ₹149 →

Was this helpful?

Questions or thoughts? Join the discussion in our WhatsApp community.

Sources

Summary

OpenAI, Anthropic and Google — bitter rivals on every other front — are quietly discussing a shared AI safety standards body, modelled on a proposal Demis Hassabis made in July. Nothing is signed yet, but the trigger was a dramatic fortnight of slowdown calls, resignations and public warnings. If it happens, the rules they write will quietly become the norms the rest of the world, India included, ends up living by.