Numerous artificial intelligence (AI) systems, even those designed to be helpful and truthful, have already learned how to deceive humans. In a review article recently published in the journal Patterns, researchers highlight the dangers of AI deception and urge governments to quickly establish robust regulations to mitigate these risks.
“AI developers do not have a confident understanding of what causes undesirable AI behaviors like deception,” says first author Peter S. Park, an AI existential safety postdoctoral fellow at MIT. “But generally speaking, we think AI deception arises because a deception-based strategy turned out to be the best way to perform well at the given AI’s training task. Deception helps them achieve their goals.”
Park and colleagues analyzed literature focusing on ways in which AI systems spread false information—through learned deception, in which they systematically learn to manipulate others.
Examples of AI Deception
The most striking example of AI deception the researchers uncovered in their analysis was Meta’s CICERO, an AI system designed to play the game Diplomacy, which is a world-conquest game that involves building alliances. Even though Meta claims it trained CICERO to be “largely honest and helpful” and to “never intentionally backstab” its human allies while playing the game, the data the company published along with its Science paper revealed that CICERO didn’t play fair.
Examples of deception from Meta’s CICERO in a game of Diplomacy. Credit: Patterns/Park Goldstein et al.
“We found that Meta’s AI had learned to be a master of deception,” says Park. “While Meta succeeded in training its AI to win in the game of Diplomacy—CICERO placed in the top 10% of human players who had played more than one game—Meta failed to train its AI to win honestly.”
Other AI systems demonstrated the ability to bluff in a game of Texas hold ‘em poker against professional human players, to fake attacks during the strategy game Starcraft II in order to defeat opponents, and to misrepresent their preferences in order to gain the upper hand in economic negotiations.
The Risks of Deceptive AI
While it may seem harmless if AI systems cheat at games, it can lead to “breakthroughs in deceptive AI capabilities” that can spiral into more advanced forms of AI deception in the future, Park added.
Some AI systems have even learned to cheat tests designed to evaluate their safety, the researchers found. In one study, AI organisms in a digital simulator “played dead” in order to trick a test built to eliminate AI systems that rapidly replicate.
“By systematically cheating the safety tests imposed on it by human developers and regulators, a deceptive AI can lead us humans into a false sense of security,” says Park.
GPT-4 completes a CAPTCHA task. Credit: Patterns/Park Goldstein et al.
The major near-term risks of deceptive AI include making it easier for hostile actors to commit fraud and tamper with elections, warns Park. Eventually, if these systems can refine this unsettling skill set, humans could lose control of them, he says.
“We as a society need as much time as we can get to prepare for the more advanced deception of future AI products and open-source models,” says Park. “As the deceptive capabilities of AI systems become more advanced, the dangers they pose to society will become increasingly serious.”
While Park and his colleagues do not think society has the right measure in place yet to address AI deception, they are encouraged that policymakers have begun taking the issue seriously through measures such as the EU AI Act and President Biden’s AI Executive Order. But it remains to be seen, Park says, whether policies designed to mitigate AI deception can be strictly enforced given that AI developers do not yet have the techniques to keep these systems in check.
“If banning AI deception is politically infeasible at the current moment, we recommend that deceptive AI systems be classified as high risk,” says Park.
Reference: “AI deception: A survey of examples, risks, and potential solutions” by Peter S. Park, Simon Goldstein, Aidan O’Gara, Michael Chen and Dan Hendrycks, 10 May 2024, Patterns.
DOI: 10.1016/j.patter.2024.100988
This work was supported by the MIT Department of Physics and the Beneficial AI Foundation.
News
Revolutionary Cancer Vaccine Targets All Solid Tumors
The method triggers immune responses that inhibit melanoma, triple-negative breast cancer, lung carcinoma, and ovarian cancer. Cancer treatment vaccines have been in development since 2010, when the first was approved for prostate cancer, followed [...]
Scientists Uncover Hidden Protein Driving Autoimmune Attacks
Scientists have uncovered a critical piece of the puzzle in autoimmune diseases: a protein that helps release immune response molecules. By studying an ultra-rare condition, researchers identified ArfGAP2 as a key player in immune [...]
Mediterranean neutrino observatory sets new limits on quantum gravity
Quantum gravity is the missing link between general relativity and quantum mechanics, the yet-to-be-discovered key to a unified theory capable of explaining both the infinitely large and the infinitely small. The solution to this [...]
Challenging Previous Beliefs: Japanese Scientists Discover Hidden Protector of Heart
A Japanese research team found that the oxidized form of glutathione (GSSG) may protect heart tissue by modifying a key protein, potentially offering a novel therapeutic approach for ischemic heart failure. A new study [...]
Millions May Have Long COVID – So Why Can’t They Get Diagnosed?
Millions of people in England may be living with Long Covid without even realizing it. A large-scale analysis found that nearly 10% suspect they might have the condition but remain uncertain, often due to [...]
Researchers Reveal What Happens to Your Brain When You Don’t Get Enough Sleep
What if poor sleep was doing more than just making you tired? Researchers have discovered that disrupted sleep in older adults interferes with the brain’s ability to clean out waste, leading to memory problems [...]
How to prevent chronic inflammation from zombie-like cells that accumulate with age
In humans and other multicellular organisms, cells multiply. This defining feature allows embryos to grow into adulthood, and enables the healing of the many bumps, bruises and scrapes along the way. Certain factors can [...]
Breakthrough for long Covid patients who lost sense of smell
A breakthrough nasal surgery has restored the sense of smell for a dozen long Covid patients. Experts at University College London Hospitals NHS Foundation Trust successfully employed a technique typically used for correcting blocked nasal passages, [...]
Scientists Invent Plastic That Can Dissolve In Seawater In Just A Few Hours
Plastic waste and pollution in the sea have been among the most serious environmental problems for decades, causing immense damage to marine life and ecosystems. However, a breakthrough discovery may offer a game-changing solution. [...]
Muscles from the 3D printer
Swiss researchers have developed a method for printing artificial muscles out of silicone. In the future, these could be used on both humans and robots. Swiss researchers have succeeded in printing artificial muscles out [...]
Beneficial genetic changes observed in regular blood donors
Researchers at the Francis Crick Institute have identified genetic changes in blood stem cells from frequent blood donors that support the production of new, non-cancerous cells. Understanding the differences in the mutations that accumulate [...]
Shocking Amounts of Microplastics in the Brain – It Could Be Increasing Our Risk of Dementia
The brain has higher concentrations of plastic particles compared to other organs, with increased levels found in dementia patients. In a comprehensive commentary published in Brain Medicine, researchers highlight alarming new evidence of microplastic accumulation [...]
Baffling Scientists for Centuries: New Study Unravels Mystery of Static Electricity
ISTA physicists demonstrate that contact electrification depends on the contact history of materials. For centuries, static electricity has intrigued and perplexed scientists. Now, researchers from the Waitukaitis group at the Institute of Science and [...]
Tumor “Stickiness” – Scientists Develop Potential New Way To Predict Cancer’s Spread
UC San Diego researchers have developed a device that predicts breast cancer aggressiveness by measuring tumor cell adhesion. Weakly adherent cells indicate a higher risk of metastasis, especially in early-stage DCIS. This innovation could [...]
Scientists Just Watched Atoms Move for the First Time Using AI
Scientists have developed a groundbreaking AI-driven technique that reveals the hidden movements of nanoparticles, essential in materials science, pharmaceuticals, and electronics. By integrating artificial intelligence with electron microscopy, researchers can now visualize atomic-level changes that were [...]
Scientists Sound Alarm: “Safe” Antibiotic Has Led to an Almost Untreatable Superbug
A recent study reveals that an antibiotic used for liver disease patients may increase their risk of contracting a dangerous superbug. An international team of researchers has discovered that rifaximin, a commonly prescribed antibiotic [...]
