This week, many of the tech world’s glitterati gathered in Lisbon for Web Summit, a sprawling conference showcasing everything from dancing robots to the influencer economy.
In the pavilions – warehouse-sized rooms chock full of stages, booths and people networking – the phrase “agentic AI” was everywhere.
There were AI agents that hung around your neck in jewellery, software to build agents into your workflows and more than 20 panel discussions on the topic.
Agentic AI is essentially artificial intelligence that can do specific tasks by itself, like book your flights or order an Uber or help a customer.
It’s the industry’s current buzzword and has even crept into the real world, with the Daily Mail listing “agentic” as an ‘in’ word for Gen Z last week.
But AI agents aren’t new. In fact, Babak Hodjat, now chief AI officer at Cognizant, invented the technology behind one of the most famous AI agents, Siri, in the 1990s.
“Back then, the fact that Siri itself was multi-agentic was a detail that we didn’t even talk about – but it was,” he told Sky News from Lisbon.
A third of daily music uploads are AI-generated and 97% of people can’t tell the difference, says report
New law could help tackle AI-generated child abuse at source, says watchdog
Ex-Starmer adviser’s start-ups backer nears £1bn valuation
“Historically, the first person that talked about something like an agent was Alan Turing.”
New or not, AI agents are thought to come with even more risks than general-purpose AI, because they interact with and modify real-world scenarios.
Please use Chrome browser for a more accessible video player
The risks that come with AI, like bias in its data or unforeseen circumstances in how it interacts with humans, are magnified by agentic AI because it interacts with the world by itself
“Agentic AI introduces new risks and challenges,” wrote the IBM Responsible Technology Board in their 2025 report on the technology.
“For example, one new emerging risk involves data bias: an AI agent might modify a dataset or database in a way that introduces bias.
“Here, the AI agent takes an action that potentially impacts the world and could be irreversible if the introduced bias scales undetected.”
But for Mr Hodjat, its not AI agents we need to worry about.
“People are over-trusting [AI] and taking their responses on face value without digging in and making sure that it’s not just some hallucination that’s coming up.
“It is incumbent upon all of us to learn what the boundaries are, the art of the possible, where we can trust these systems and where we cannot, and educate not just ourselves, but also our children.”
His warning will feel familiar, particularly in Europe, where there’s an increased wariness around AI compared to the US.
But have we become too cautious when it comes to AI – at the risk of a far more existential threat in the future?
Jarek Kutylowski, chief executive of German AI language giant DeepL, certainly thinks so.
This year, the EU AI Act came into force, strict regulations about how companies can and can’t use AI.
In the UK, companies are governed by existing legislation like GDPR and there’s uncertainty about how strict our rules will be in the future.
When asked if we needed to slow down AI innovation in order to put stricter regulations in place, Mr Kutylowski said it was a question worth grappling with… but in Europe, we are taking it too far.
Read more from science and technology:
NASA cancels space launch
Jeff Bezos’s rocket lands on Earth
New law could help tackle AI-generated child abuse at source
“Looking at the apparent risks is easy, looking at the risks like what are we going to miss out on if we don’t have the technology, if we are not successful enough in adopting that technology, that is probably the bigger risk,” said Mr Kutylowski.
“I see definitely a much larger risk in Europe being left behind in the AI race.”
“You won’t see it until we start falling behind and until our economies cannot capitalise on those productivity gains that maybe other parts of the world will see.
“I do not believe personally that technological progress can be stopped in any way, so it is more of a question of ‘how do we pragmatically embrace what is coming ahead?”
