Post-Doctoral Research Visit F/M From AI audits to AI security: an information gain hierarchy

Contexte et atouts du poste

AI-based models are now core to a wide range of applications,
including highly critical ones. The stakes are considerable for
companies or institutions deploying them, as their training amounts to
up to a billion dollars (e.g. for the training of ChatGPT). This
clearly calls for defending them against attacks, like
copy/extraction. In parallel, institutions such as state regulators
have to ensure that these models operate according to law, in
particular with regards to possible discrimination [1]. Researchers
are then tasked to provide algorithms to auditors for assessing
important metrics regarding deployed AI-based models. In such
black-box audits, where an auditor has no access to the remotely
operated model's internals, the goal is to stealthily (i.e. with a few
queries only) estimate some metrics, such as fairness [6].
Interestingly, and this has not yet been mentioned in the literature,
this audit setup is very close to offensive information gain, from a
conceptual standpoint. Indeed, potential attackers are incentivized to
try and leak information out of deployed models [4,10]. Motivations
range from economic intelligence, obtaining implementation details, to
simply avoiding development costs by copying deployed models. An
auditor is interested in stealthy model observation [3], to avoid
disrupting the audited model by using too many queries. Identically,
an attacker also desire stealthiness, here to avoid being detected and
cut off. In particular, auditors generally want to obtain precise
property estimation, yet confined to a single feature
(e.g. male/female fairness), while attackers aim at having a global
picture of the model (for basic copy, or evading some parameter
set). Thus, there is an avenue to devise offensive methods in between
stealthy audits and global attacks, to try and leak novel model
characteristics. The ambition of our group is to bridge the gap
between these two critical setups: legal auditing and offensive
security, in the domain of modern deployed AI models. From this unique
standpoint, and from the body of work in the field of AI auditing, we
expect to find new insights for attacking and defending deployed AI
models, by finding novel angles. For instance, we proposed a unified
way to approach model fingerprinting [2] that is of interest for an
auditor to guess which model she is observing on a platform; we
conjecture that leveraging such an approach to measure the evolution
in time of such a model (does the model changes due to updates?) is of
core interest for an attacker, as she can derive what is at play at
the company hosting this model. This could provide ground for the
attacker for economic intelligence, while leaking some precious
information that has to be defended by the attacked company.

Mission confiée

• Research
• Working with Ph.D. students from the group

Principales activités

A striking remark when looking at the current types of attacks on AI
models is their quantity and apparent independence (see [10] Fig. 3):
each is treated as a separate domain. In addition to this list of
attacks, we claim that an audit may be viewed as the leak of a feature
from a production model, and must be considered as a potential threat.
In that light, clarifications in the relation between these attacks
might come from a systematic study of how they relate with regards to
the setup they operate in, versus the information gain they permit. We
propose to work on a hierarchy of attacks, that will uncover the
smallest attacks (in terms of assumptions and scope) and how they
might be composed into larger attacks, and so on. This hierarchy will
reveal unexplored configurations, where several simple attacks will be
combined to build richer attacks. This hierarchy will provide the
missing link between audits and AI security, bridging the two in a
formal way. The postdoc candidate will leverage algorithmic
background, to devise a hierarchy, in a parallel to the Herlihy
hierarchy in algorithms. We intend to use the notion of
"distinguishability" [14] as a hierarchy backbone (to assess if an
attack leaks data permitting strong or weak distinguishability of
models). In particular, the field of "property testing" will be
related to this hierarchy.

# References

[1] Le Merrer, E., Pons, R., & Tredan, G. . Algorithmic audits of algorithms, and the law. AI and Ethics, 4,
1365-1375.
[2] Godinot, A., Le Merrer, E., Penzo, C., Taïani, F., & Tredan, G. . Queries, Representation & Detection: The
Next 100 Model Fingerprinting Schemes. In AAAI.
[3] Le Merrer, E., & Tredan, G. Remote explainability faces the bouncer problem. Nature machine intelligence,
2, 529-539.
[4] Maho, T., Furon, T., & Le Merrer, E. . Surfree: a fast surrogate-free black-box attack. In CVPR.
[5] Godinot, A., Le Merrer, E., Tredan, G., Penzo, C., & Taïani, F. . Under manipulations, are some AI models
harder to audit?. In IEEE Conference on Secure and Trustworthy Machine Learning.
[6] de Vos, M., Dhasade, A., Garcia Bourrée, J., Kermarrec, A. M., Le Merrer, E., Rottembourg, B., & Tredan, G. .
Fairness auditing with multi-agent collaboration. In ECAI.
[7] Le Merrer, E., Perez, P., & Tredan, G. . Adversarial frontier stitching for remote neural network
watermarking. Neural Computing and Applications, 32, 9233-9244.
[8] Le Merrer, E., Morgan, B., & Tredan, G. . Setting the record straighter on shadow banning. In INFOCOM.
[9] Maho, T., Furon, T., & Le Merrer, E. . Randomized smoothing under attack: How good is it in practice?. In
ICASSP.
[10]Ma et al., « Safety at Scale: A Comprehensive Survey of Large Model Safety». arXiv:2502.05206v3
[11]Yan, T., & Zhang, C. . Active fairness auditing. In ICML.
[12]Apruzzese, G., Anderson, H. S., Dambra, S., Freeman, D., Pierazzi, F., & Roundy, K. . “Real attackers don't
compute gradients”: bridging the gap between adversarial ml research and practice. In 2023 IEEE conference on
secure and trustworthy machine learning.
[13]Fukuchi, K., Hara, S., & Maehara, T. . Faking fairness via stealthily biased sampling. In AAAI.
[14]Attiya, H., & Rajsbaum, S. . Indistinguishability. Communications of the ACM, 63, 90-99.
[15]ANSSI . Security recommandations for a generative AI system. ANSSI-PA-102.

Compétences

• Advanced machine learning background, and theory of machine learning
• Python coding skills for experiments (if required)
• A good publication track record is mandatory
• Fluency in English is mandatory

Avantages

• Subsidized meals
• Partial reimbursement of public transport costs
• Leave: 7 weeks of annual leave + 10 extra days off due to RTT (statutory reduction in working hours) + possibility of exceptional leave (sick children, moving home, etc.)
• Possibility of teleworking (after 6 months of employment) and flexible organization of working hours
• Professional equipment available (videoconferencing, loan of computer equipment, etc.)
• Social, cultural and sports events and activities
• Access to vocational training
• Social security coverage

Rémunération

Monthly gross salary amounting to 2788 euros