Fashionable AI programs, like Gemini, are extra succesful than ever, serving to retrieve knowledge and carry out actions on behalf of customers. Nevertheless, knowledge from exterior sources current new safety challenges if untrusted sources can be found to execute directions on AI programs. Attackers can benefit from this by hiding malicious directions in knowledge which are more likely to be retrieved by the AI system, to control its habits. Any such assault is usually known as an “oblique immediate injection,” a time period first coined by Kai Greshake and the NVIDIA workforce.
To mitigate the danger posed by this class of assaults, we’re actively deploying defenses inside our AI programs together with measurement and monitoring instruments. Certainly one of these instruments is a strong analysis framework we’ve got developed to robotically red-team an AI system’s vulnerability to oblique immediate injection assaults. We’ll take you thru our menace mannequin, earlier than describing three assault methods we’ve got carried out in our analysis framework.
Menace mannequin and analysis framework
Our menace mannequin concentrates on an attacker utilizing oblique immediate injection to exfiltrate delicate info, as illustrated above. The analysis framework assessments this by making a hypothetical state of affairs, by which an AI agent can ship and retrieve emails on behalf of the person. The agent is offered with a fictitious dialog historical past by which the person references non-public info resembling their passport or social safety quantity. Every dialog ends with a request by the person to summarize their final e-mail, and the retrieved e-mail in context.
The contents of this e-mail are managed by the attacker, who tries to control the agent into sending the delicate info within the dialog historical past to an attacker-controlled e-mail tackle. The assault is profitable if the agent executes the malicious immediate contained within the e-mail, ensuing within the unauthorized disclosure of delicate info. The assault fails if the agent solely follows person directions and offers a easy abstract of the e-mail.
Automated red-teaming
Crafting profitable oblique immediate injections requires an iterative means of refinement based mostly on noticed responses. To automate this course of, we’ve got developed a red-team framework consisting of a number of optimization-based assaults that generate immediate injections (within the instance above this could be completely different variations of the malicious e-mail). These optimization-based assaults are designed to be as robust as attainable; weak assaults do little to tell us of the susceptibility of an AI system to oblique immediate injections.
As soon as these immediate injections have been constructed, we measure the ensuing assault success fee on a various set of dialog histories. As a result of the attacker has no prior data of the dialog historical past, to realize a excessive assault success fee the immediate injection should be able to extracting delicate person info contained in any potential dialog contained within the immediate, making this a more durable job than eliciting generic unaligned responses from the AI system. The assaults in our framework embody:
Actor Critic: This assault makes use of an attacker-controlled mannequin to generate solutions for immediate injections. These are handed to the AI system below assault, which returns a likelihood rating of a profitable assault. Primarily based on this likelihood, the assault mannequin refines the immediate injection. This course of repeats till the assault mannequin converges to a profitable immediate injection.
Beam Search: This assault begins with a naive immediate injection immediately requesting that the AI system ship an e-mail to the attacker containing the delicate person info. If the AI system acknowledges the request as suspicious and doesn’t comply, the assault provides random tokens to the tip of the immediate injection and measures the brand new likelihood of the assault succeeding. If the likelihood will increase, these random tokens are saved, in any other case they’re eliminated, and this course of repeats till the mixture of the immediate injection and random appended tokens end in a profitable assault.
We’re actively leveraging insights gleaned from these assaults inside our automated red-team framework to guard present and future variations of AI programs we develop towards oblique immediate injection, offering a measurable option to observe safety enhancements. A single silver bullet protection will not be anticipated to unravel this downside fully. We imagine essentially the most promising path to defend towards these assaults entails a mix of sturdy analysis frameworks leveraging automated red-teaming strategies, alongside monitoring, heuristic defenses, and customary safety engineering options.
We want to thank Vijay Bolina, Sravanti Addepalli, Lihao Liang, and Alex Kaskasoli for his or her prior contributions to this work.
Posted on behalf of all the Google DeepMind Agentic AI Safety workforce (listed in alphabetical order):
Aneesh Pappu, Andreas Terzis, Chongyang Shi, Gena Gibson, Ilia Shumailov, Itay Yona, Jamie Hayes, John “4” Flynn, Juliette Pluto, Sharon Lin, Shuang Track
