According to Alan Winfield, a professor at the University of the West of England, scientists are spending too much time developing artificial intelligence and too little on robot safety.

“We’re rapidly approaching the time when new robots should undergo tests, similar to ethical and clinical trials for new drugs, before they can be introduced,” he said.

The scientists who presented their findings at the International Joint Conference for Artificial Intelligence in Pasadena, California, last month fear that nightmare scenarios, which have until now been limited to science fiction films, such as the Terminator series, The Matrix, 2001: A Space Odyssey and Minority Report, could come true.

Robotic unmanned predator drones, for example, which can seek out and kill human targets, have already moved out of the movie theatres and into the theatre of war in Afghanistan and Iraq. While at present controlled by human operators, they are moving towards more autonomous control.

They could also soon be found on the streets. Samsung, the South Korean electronics company, has developed autonomous sentry robots to serve as armed border guards. They have “shoot-to-kill” capability.

Noel Sharkey, professor of artificial intelligence and robotics at Sheffield University, warned that such robots could soon be used for policing, for example during riots such as those seen in London at the recent G20 summit. “Is this a good thing?” he asked.

Scientists are particularly worried about the way the latest, highly sophisticated artificially intelligent products perform human-like functions.

Japanese consumers can already buy robots that “learn” their owner’s behaviour, can open the front door and even find electrical outlets and recharge themselves so they never stop working.

One high-tech US firm is working on robotic nurses, dubbed “nursebots”, that interact with patients to simulate empathy. Critics told the conference that, at best, this could be dehumanising; at worst, something could go wrong with the programming.

The scientists dismissed as fanciful fears about “singularity” — the term used to describe the point where robots have become so intelligent they are able to build ever more capable versions of themselves without further input from mankind.

The conference was nevertheless told that new artificial intelligence viruses are helping criminals to steal people’s identities. Criminals are working on viruses that are planted in mobile phones and “copy” users’ voices. After stealing the voice, criminals can masquerade as a victim on the phone or circumvent speech recognition security systems.

Another kind of smartphone virus silently monitors text messages, e-mail, voice, diary and bank details. The virus then uses the information to impersonate people online, with little or no external guidance from the thieves. The researchers warned that many of the new viruses defy extermination, reaching what one speaker called “the cockroach stage”.

Some speakers called for researchers to adopt the “three laws” of robotics created by Isaac Asimov, the science fiction author, that are designed to protect humanity from machines with their own agenda. Each robot, Asimov said, must be programmed never to kill or injure a human or, through inaction, allow a human to suffer. A robot must obey human orders, unless this contravenes the first law. A robot must protect itself, unless it contravenes either of the first two laws.

While many scientists fear artificial intelligence could run amok, some argue that ultrasmart machines will instead offer huge advances in life extension and wealth creation.

Some pointed out that artificial intelligence was already helping us in complex, sometimes life-and-death situations. Poseidon Technologies, the French firm, sells artificial intelligence systems that help lifeguards identify when a person is drowning in a swimming pool. Microsoft’s Clearflow system helps drivers to pick the best route by analysing traffic behaviour; and artificial intelligence systems are making cars safer, reducing road accidents.