Geoffrey Hinton has sent out a warning beacon to the rest of humanity. He fears AI could wipe out our existence. And he should know. Hinton helped build the technology.

The Nobel Prize-winning computer scientist doesn’t mince his words and explicitly says that “tech bros” are taking the wrong approach. And if “the godfather of AI” expresses doubts about the future of humankind, surely we should pay attention?

Don’t get us wrong. Since the inception of artificial intelligence, we’ve seen extraordinary innovations. Intelligent systems and machine learning are redefining entire ecosystems. Yet, they come with unprecedented risks, warns the World Economic Forum.

If the technology goes wrong, who or what is to blame?

Why is Accountability So Hard?

Technically, AI doesn’t make decisions as we do. Even its creators can’t fully explain how a complex model arrived at a specific output.

That elicits a kind of “fog” around responsibility. If an AI misdiagnoses a patient or wrongly rejects a loan application, who bears the blame? The engineer who designed the model? The company that deployed it? The user who trusted the result? Or no one at all?

The lack of clear accountability stems partly from the fact that AI decisions are not always explainable. And that in itself is scary.

When the internal logic of a system is basically a black box, it’s difficult to trace what went wrong and why, and harder to assign fault.

AI and the Law: A Hazy Patchwork

In many countries, existing laws weren’t written with AI in mind. Legal concepts such as negligence and product liability assume a clear chain of human decision-making. What happens when software plays a role in a harmful outcome?

For example, Gianaris Trial Lawyers explains that personal injury law requires showing that someone had a duty of care, neglected that duty, and caused harm as a result. If an AI chatbot gives misleading advice that leads to a financial loss, was there a duty of care? Was it breached? And if so, by whom?

These questions become more complicated in the AI context.

There are cases under common law where courts have decided organizations should be held liable for faulty AI when harm occurs. They treat the system’s actions as those of the business itself. But this is not yet widespread or consistent.

Corporate Negligence

One of the biggest problems isn’t the AI itself; it’s how companies handle responsibility.

Some organizations focus on rushing new systems to market without building in accountability from the start. That’s a form of corporate negligence.

Only when the harm is done do executives sometimes treat responsibility like a mystery to be solved later. Instead, it’s supposed to be a principle already built in.

Liability isn’t only a legal issue, but an ethical and managerial one. Senior leadership, boards, and risk officers must ensure systems are safe before they’re deployed.

AI governance frameworks shouldn’t merely check a box. Boards need to understand both the technical risks and how they translate to real-world harms.

Accountability in Practice

Some experts argue that accountability for AI shouldn’t rest with a single actor alone. It should be shared across a system of stakeholders:

• Developers who build the algorithms and choose how they learn from data.
  
• Deployers who decide where to apply the systems and how they’re supervised.
  
• Corporate leaders who set risk tolerance and governance policies.
  
• End users who interpret and act on AI outputs.

Separating these responsibilities helps avoid blanket blame and creates clarity about who is responsible for each part of the process.

Real Harm, Real Reflection

All of this might sound abstract. The consequences are very real.

AI systems have already shown their ugly side, mirroring societal biases or amplifying harmful outcomes when not carefully governed. Some commentators describe AI as reflecting our worst selves when we outsource judgment without proper controls.

Lawsuits and regulatory fines aside, it all comes down to trust. If people can’t trust that technology will behave safely, innovation suffers.

What Does AI Accountability Look Like?

Rather than vague promises or nice-sounding ethics statements, credible accountability involves:

• Clever governance

• Transparency

• Shared responsibility

• Legal preparedness

AI’s brilliance is undeniable. But brilliance without accountability is dangerous.

We build powerful tools to help people. When those tools fail, we can’t shrug and say, “Well, it was just a machine.”

Behind every algorithm are humans who made design choices and leaders who decided it was ready. If we want AI to benefit humanity, we must put responsibility on the table before something goes wrong.