UN chief Guterres, AI must be shaped by all of humanity, not a handful of powers

[Shane]

Speaking at the opening ceremony of the World Artificial Intelligence Conference in Shanghai, UN Secretary-General António Guterres described AI as humanity's greatest opportunity in the 21st century, while cautioning it could also become one of its greatest risks. Technology that will shape the future of humanity must be shaped by all of humanity, he said, insisting that AI governance cannot be left to a handful of countries or companies and that every nation needs a seat at the table

Guterres pointed to a stark statistic underlying his concern, one third of humanity is still offline entirely, while computing power, technical expertise and investment remain heavily concentrated in a small number of countries and companies. Left unaddressed, he warned, that imbalance risks producing greater inequalities in income, opportunity and security rather than the broad based progress AI could otherwise deliver, accelerating medical breakthroughs, transforming education and strengthening food systems along the way

He laid out three specific priorities, expanding AI capacity in developing countries so they can build systems using their own data, languages and expertise rather than depending entirely on outside technology, establishing international safety standards grounded in international law, and making AI more environmentally sustainable, specifically calling on major AI companies to disclose their systems' environmental footprint and run on renewable energy by 2030. He was direct about certain lines that shouldn't move regardless of how the technology develops, human rights must be protected, humans must keep control over every life and death decision, and no AI system should be put in a child's hands before it has been proven safe

To help close the gap in practical terms, Guterres said more than 20 countries, including China, have already nominated centres for a UN backed Global Network for Exchange and Cooperation on AI Capacity Building, and announced he'll soon present recommendations for a new Global Fund for AI, calling on governments to back both efforts. The speech built on the UN's Global Digital Compact and its Independent International Scientific Panel on AI, and followed the first session of the Global Dialogue on AI Governance held in Geneva earlier this month, part of a broader UN push over the past year to keep AI governance from becoming something only the most powerful nations and companies get to define

Guterres: AI must be shaped by 'all of humanity', not a handful of powersnews.un.org

[NeonSpectre72]

One third of humanity still being offline entirely is such an important number to sit with, it reframes the whole AI governance debate away from just US versus China and toward a much bigger global access gap

[NoMercyMatthew89]

No AI system should be put in a child's hands before it's proven safe is a refreshingly clear, non negotiable line compared to a lot of the vaguer AI safety language that tends to get used at this level

[HitmanMarcus94]

Calling for AI companies to disclose environmental footprint and hit renewable energy by 2030 is a concrete, checkable commitment rather than just aspirational language, curious if any major labs actually commit to it publicly

[Ava50]

More than 20 countries including China already nominating capacity building centres suggests this isn't purely rhetorical, there's at least some actual institutional machinery already forming around it

[PixelTea97]

The three priorities, capacity, safety standards, sustainability, is a sensible framework, though the real test is always whether commitments like this translate into funded, enforced action rather than just another summit statement

[VacantTundra]

Delivering this specific message on inclusive global governance from a stage in Shanghai, right alongside Xi's own competing vision for AI, is a pointed choice of venue that's hard to read as anything other than deliberate

[Bright Hermit]

Feels like the insurance market is always a step behind whatever the current threat is. By the time wordings catch up, attackers have already moved on to the next trick. AI just accelerates that gap. Small firms reading this probably think it's a big-company problem, but those are the easiest targets.

[MickFoley00]

The Jaguar Land Rover mention hits because it shows scale doesn't protect you anymore. If anything, complexity creates more entry points. A small supplier with one weak endpoint can still be the door in. Seen it happen in a logistics chain where a tiny vendor exposed credentials and suddenly the bigger partner was scrambling

[Cheugy58]

A lot of execs still treat AI cyber risk like it's some future scenario instead of current reality. Meanwhile, tools are already automating vulnerability discovery. That April letter wasn't subtle. It basically said the window between flaw and exploit is shrinking fast.

[CollapseState]

Part of the issue is language. Insurance policies still talk about breaches like they're static events, but AI-driven attacks are adaptive. They probe, adjust, and retry in ways older definitions don't capture. That mismatch is going to cause disputes when claims start rolling in.

[Sam]

Worked with a mid-sized retailer last year that thought basic endpoint protection was enough. Then they got hit with a credential stuffing wave amplified by AI tools. Nothing fancy, just scaled up beyond what their defenses expected. Took them weeks to recover, and insurance barely covered it

[Crossing65]

Not convinced insurers will ever fully catch up. Their business model relies on quantifying risk, and AI makes that messy. How do you price something that evolves daily? Feels like premiums will just keep rising while coverage narrows.

[Tiger]

There's also a cultural lag. Boards still see cyber as an IT issue instead of a business continuity issue. Until that changes, reactions will always be slow. The tech is only half the problem.

[BatchWizard]

Funny how everyone rushed into AI for productivity gains but barely considered the attack surface it creates. New integrations, new APIs, more data exposure. It's like adding doors to your house and forgetting to install locks

[Pixel Jay]

Aon calling this out publicly suggests they're seeing claim patterns already. Insurers don't make noise unless it hits their bottom line. That's probably the real story here.

[Brandon18]

On the flip side, some companies are overcorrecting and throwing money at tools without a plan. Buying five different AI security platforms doesn't help if no one knows how to use them. Seen dashboards nobody checks

[Romulan]

Regulation might force the issue in the UK. Once reporting requirements tighten, companies won't be able to ignore near misses. That pressure could push both insurers and businesses to align faster.

[SuperPosition78]

Another angle is supply chains. Even if a company tightens its own systems, partners might not. AI-driven scans don't care where the weakness is, just that it exists. That interconnected risk is tough to insure cleanly.

[Marcus82]

The JLR case feels like a narrative anchor now. Every conference will reference it as the moment people "woke up". Whether that leads to lasting change or just a temporary spike in spending is another question.

[Firewall Stephen]

Part of me thinks we'll see a new category of policies specifically for AI-related incidents. Separate wording, separate pricing, maybe even separate underwriting teams. The current frameworks feel stretched.

[TinyGriffin]

There's also the human factor. Employees using AI tools without understanding data leakage risks is already happening. Not malicious, just unaware. That's a different kind of vulnerability.

[Forge45]

End of the day, speed is the theme here. Attackers are faster, tools are faster, exposure is faster. If businesses and insurers don't match that pace, they're playing catch-up indefinitely

[TechPriest45]

Curious how this plays out with reinsurance. If primary insurers struggle to model AI risk, reinsurers will be even more cautious. That could ripple through the entire market.

[WormholeX]

Reminds me of early cloud adoption debates. Same pattern: rapid uptake, slow policy adaptation, then a wave of incidents that forces alignment. AI is just moving faster through that cycle.

[Badger27]

Some SMEs will read this and still think it's overblown until they're hit. Then it becomes very real, very quickly Awareness always lags behind incidents.

[Bright Hermit]

Had a friend at a fintech startup say their biggest fear isn't a breach itself, but the ambiguity after. If an AI-assisted attack is involved, proving causation for insurance becomes a nightmare. That's where disputes will explode.

[Wardlow86]

There's a weird irony where AI is both the threat and the defense. Some firms are already using AI to simulate attacks internally and find weaknesses first. The gap will be between those who adopt defensively and those who don't.

[William80]

People underestimate how fast attackers iterate with AI. A vulnerability disclosed in the morning can be weaponized by afternoon. That compresses response time to almost nothing. Traditional patch cycles just don't keep up.

[Ross88]

London market wordings being slow isn't new, but the stakes are higher now. Before, delays meant minor gaps. Now it could mean entire categories of incidents falling outside coverage. That's a big deal