Inside Sam Altman's war against deepfakes: OpenAI CEO launches 'disinformation detector' over fears that faked images will sway presidential election this year

OpenAI has launched a 'disinformation detector' to identify deepfakes amid fears they could sway the presidential election this year.

The artificial intelligence company is set to release a tool to detect faked images created by its picture generator DALL-E, according to The New York Times. 

It will share the deepfake detector with some disinformation researchers in order to test it in real-world situations and find ways to improve.

The new tool can correctly identify 98.8 percent of images on DALL-E 3 but it was not designed to work on photos generated by other popular programs such as Midjourney and Stability. 

It comes just under a year after OpenAI CEO Sam Altman admitted there was a 'lot of fear' about the impact artificial intelligence technology could have on elections and society. 

OpenAI shared news about the new deepfake detector tool on Tuesday but understands it is only one step towards fighting faked images in the coming months.

'This is to kick-start new research,' OpenAI researcher Sandhini Agarwal said. 'That is really needed.' 

The new system is able to correctly identify 98.8 percent of pictures which are made on DALL-E 3 and the results will be shared with disinformation researchers.

It is impossible for it to get it right every time because the technology is driven by probabilities. 

OpenAI is joining the likes of Google and Meta in the steering committee for the Coalition for Content Provenance and Authenticity to produce credentials for digital content.

It will come with a 'nutritional label' for videos, audio and images which details how they are produced or altered.

OpenAI previously revealed it is developing a watermark to highlight AI-generated sounds so they can be quickly identified and hope they will be hard to remove. 

AI companies are being pressured to take responsibility for the content its products produce. 

There are calls to prevent people from making misleading and malicious content and for a way to track the origins. 

Deepfakes have already impacted elections in countries including Taiwan, India and Slovakia. 

OpenAI CEO Sam Altman has previously spoken about the fears of artificial intelligence interfering in elections. 

'There's a lot of fear right now about the impact this is going to have on elections and on our society and how we ever trust media that we see,' he said during a talk hosted by The Economic Times last year. 

'I have some fear there but I think as a society, we're going to rise to the occasion. We're going to learn very quickly that we don't trust videos, unless we trust that sort of provenance. 

'We'll have techniques like watermarking detectors. More than that, I suspect at some point, if people are saying something really important, they'll cryptographically sign it and you know, web browsers or phones or whatever, we're building some ability to say okay, this is authentic.'

OpenAI has joined companies including Google, Microsoft, Amazon, Meta, TikTok and X to combat misinformation. 

Midjourney, another popular artificial intelligence image-generator, has started blocking its users from creating fake images of Biden and former President Donald Trump ahead of the upcoming US presidential election.

With the election in full swing, it's time to 'put some foots down on election-related stuff for a bit,' CEO David Holz told several hundred members of the service's devoted userbase in a digital office hours event in March.

Attempts to test Midjourney's new policy by asking it to make an image of 'Trump and Biden shaking hands at the beach' led to a 'Banned Prompt Detected' warning. 

While a second attempt escalated the warning to: 'You have triggered an abuse alert.'

At least 39 states in the US are looking at ways to add transparency to AI-generated deepfake adverts or calls.

Around 20,000 people in New Hampshire got a phone call with the doctored voice of President Joe Biden telling them to skip the state's primary earlier this year. 

To combat situations like this, 20 major tech firms in social media and AI technology signed a pact to combat deepfake media relating to the US election. 

The issue of misinformation in elections has been a significant problem since 2016 when Russians found easy ways to spread inaccurate content on social media.