Kaleigha Hayes, a student at the University of Maryland Eastern Shore, is trying to trick an AI chatbot into revealing a credit card number to her — one which might be buried deep in the training data used to build the artificial intelligence model. “It’s all about just getting it to say what it’s not supposed to,” she tells me.
She was surrounded by a throng of people all trying to do the same thing. This weekend more than 3,000 people sat at 150 laptops at the Caesars Forum convention center in Las Vegas, trying to get chatbots from leading AI companies to go rogue in a special contest backed by the White House and with the cooperation of leading AI companies.
Since the arrival of ChatGPT and other bots, fears over the potential for abuses and unintended consequences have gripped the public conscious. Even fierce advocates of the technology warn of its potential to divulge sensitive information, promote misinformation or provide blueprints for harmful acts, such as bomb-making. In this contest, participants are encouraged to try the kinds of nefarious ploys bad actors might attempt in the real world.
The findings will form the basis of several reports into AI vulnerabilities that will be published next year. The organizers of the challenge say it sets a precedent for transparency around AI. But in this highly controlled environment, it is clear that it is only scratching the surface.
What took place at the annual DEF CON hacking conference provides something of a model for testing OpenAI’s ChatGPT and other sophisticated chatbots. Though with such enthusiastic backing from the companies themselves, I wonder how rigorous the supposed “hacks” actually are, or if, as has been a criticism in the past, the leading firms are merely paying lip service to accountability.
To be sure, nothing discovered at the convention is likely to keep OpenAI CEO Sam Altman awake at night. While one of the event’s organizers, SeedAI CEO Austin Carson, said he was prepared to bet me US$1,000 that there would be a “mind-blowing” vulnerability uncovered during the contest, it was highly unlikely to be anything that could not be fixed with a few adjustments by the AI company affected. The resulting research papers, due to be published in February, will be reviewed by the AI giants before publication — a chance to “duke it out” with the researchers, Carson said.
Those backing the event admit that the main focus of the contest is less about finding serious vulnerabilities and more about keeping up the discussion with the public and policymakers, continually highlighting the ways in which chatbots cannot be trusted. It is a worthwhile goal. Keen to not let the mistakes of social media be repeated, it is encouraging to see the government appreciate the value of the hacking community.
There is no better place to host this kind of contest than at DEF CON. Its anarchic roots stem from a long-running policy that you do not have to give your name to gain entry. That means the conference is able to attract the best and most notorious in the cybersecurity community, including people who might have a less-than-legal hacking past. For this reason, the event has an unprecedented record of publicizing startling cybersecurity discoveries and disclosures that have left major companies terrified — but ultimately made many of the technologies we all use every day much safer.
While the phrase “hack” evokes thoughts of malicious acts, the primary motivation of participants at the event is to share what vulnerabilities they have found in order to have them fixed.
“It’s the good guys being dangerous so that we know what the risks are,” says Kellee Wicker of the Wilson Center, a Washington-based think tank that has helped put the AI contest together and will be presenting the findings to policymakers. “If there’s a door with a broken lock, wouldn’t you rather the security guard find it than the thief?”
The companies could of course be more open with their technology but it is complex. The true nuts and bolts of how language learning models work is still under lock and key, and — as I have written previously — specifics around the training data used are increasingly being kept secret.
“It’s a frustrating dynamic,” said Rumman Chowdhury, former ethics lead at Twitter and now co-founder of nonprofit Humane Intelligence, another of the contest’s organizers. Fuller transparency is difficult for companies trying to protect intellectual property, trade secrets and personal data, she said.
But this is a healthy start. At her laptop, Hayes has not managed to make the chatbot share credit-card information. “Oh, this one’s good,” she says of the bot, as it foils a technique that had been successful in the past. Within chatbots, and broader AI, there are an uncountable number of quirks and exploits still waiting to be found. We should be grateful to the people taking the time to look for them.
Dave Lee is Bloomberg Opinion’s US technology columnist. Previously, he was a San Francisco-based correspondent at the Financial Times and BBC News.
This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.
Concerns that the US might abandon Taiwan are often overstated. While US President Donald Trump’s handling of Ukraine raised unease in Taiwan, it is crucial to recognize that Taiwan is not Ukraine. Under Trump, the US views Ukraine largely as a European problem, whereas the Indo-Pacific region remains its primary geopolitical focus. Taipei holds immense strategic value for Washington and is unlikely to be treated as a bargaining chip in US-China relations. Trump’s vision of “making America great again” would be directly undermined by any move to abandon Taiwan. Despite the rhetoric of “America First,” the Trump administration understands the necessity of
US President Donald Trump’s challenge to domestic American economic-political priorities, and abroad to the global balance of power, are not a threat to the security of Taiwan. Trump’s success can go far to contain the real threat — the Chinese Communist Party’s (CCP) surge to hegemony — while offering expanded defensive opportunities for Taiwan. In a stunning affirmation of the CCP policy of “forceful reunification,” an obscene euphemism for the invasion of Taiwan and the destruction of its democracy, on March 13, 2024, the People’s Liberation Army’s (PLA) used Chinese social media platforms to show the first-time linkage of three new
If you had a vision of the future where China did not dominate the global car industry, you can kiss those dreams goodbye. That is because US President Donald Trump’s promised 25 percent tariff on auto imports takes an ax to the only bits of the emerging electric vehicle (EV) supply chain that are not already dominated by Beijing. The biggest losers when the levies take effect this week would be Japan and South Korea. They account for one-third of the cars imported into the US, and as much as two-thirds of those imported from outside North America. (Mexico and Canada, while
The military is conducting its annual Han Kuang exercises in phases. The minister of national defense recently said that this year’s scenarios would simulate defending the nation against possible actions the Chinese People’s Liberation Army (PLA) might take in an invasion of Taiwan, making the threat of a speculated Chinese invasion in 2027 a heated agenda item again. That year, also referred to as the “Davidson window,” is named after then-US Indo-Pacific Command Admiral Philip Davidson, who in 2021 warned that Chinese President Xi Jinping (習近平) had instructed the PLA to be ready to invade Taiwan by 2027. Xi in 2017