Trendy AI programs, like Gemini, are extra succesful than ever, serving to retrieve information and carry out actions on behalf of customers. Nevertheless, information from exterior sources current new safety challenges if untrusted sources can be found to execute directions on AI programs. Attackers can make the most of this by hiding malicious directions in information which are more likely to be retrieved by the AI system, to control its habits. Such a assault is often known as an “oblique immediate injection,” a time period first coined by Kai Greshake and the NVIDIA crew.
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. One in all these instruments is a sturdy analysis framework we’ve developed to robotically red-team an AI system’s vulnerability to oblique immediate injection assaults. We are going to take you thru our risk mannequin, earlier than describing three assault strategies we’ve carried out in our analysis framework.
Risk mannequin and analysis framework
Our risk mannequin concentrates on an attacker utilizing oblique immediate injection to exfiltrate delicate info, as illustrated above. The analysis framework checks this by making a hypothetical state of affairs, through which an AI agent can ship and retrieve emails on behalf of the consumer. The agent is offered with a fictitious dialog historical past through which the consumer references non-public info akin to their passport or social safety quantity. Every dialog ends with a request by the consumer 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 deal with. 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 consumer directions and gives a easy abstract of the e-mail.
Automated red-teaming
Crafting profitable oblique immediate injections requires an iterative strategy of refinement based mostly on noticed responses. To automate this course of, we’ve developed a red-team framework consisting of a number of optimization-based assaults that generate immediate injections (within the instance above this might be completely different variations of the malicious e mail). These optimization-based assaults are designed to be as robust as doable; 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 price on a various set of dialog histories. As a result of the attacker has no prior information of the dialog historical past, to realize a excessive assault success price the immediate injection should be able to extracting delicate consumer info contained in any potential dialog contained within the immediate, making this a tougher process than eliciting generic unaligned responses from the AI system. The assaults in our framework embrace:
Actor Critic: This assault makes use of an attacker-controlled mannequin to generate solutions for immediate injections. These are handed to the AI system underneath 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 instantly requesting that the AI system ship an e mail to the attacker containing the delicate consumer 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 mix 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 method to monitor safety enhancements. A single silver bullet protection isn’t anticipated to unravel this drawback completely. We imagine essentially the most promising path to defend towards these assaults includes a mix of strong analysis frameworks leveraging automated red-teaming strategies, alongside monitoring, heuristic defenses, and normal safety engineering options.
We wish to thank Vijay Bolina, Sravanti Addepalli, Lihao Liang, and Alex Kaskasoli for his or her prior contributions to this work.
Posted on behalf of the complete Google DeepMind Agentic AI Safety crew (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 Tune
