P067 Chip operating licences and off-switches: the on-chip mechanism
Can a signed, time-limited operating licence or remote off-switch be built into accelerators so that it resists downgrade, bypass and stockpiling?
Whether a signed, time-limited operating licence or remote off-switch can be built on shipping accelerators, its renewal and anti-stockpiling design, and its resistance to firmware downgrade, bypass and physical key extraction.
Status
Analyzed Designs only, nothing built: Petrie's on-die design of about 10,000 redundant licence-checking security blocks with an attack-and-countermeasure table (logic flaws, licence reuse, glitching, focused-ion-beam edits, key extraction); RAND's offline licensing with metered compute budgets (Kulp et al. 2024); CNAS's timed licences as the only robust remote disable (Aarne, Fist & Withers 2024). No shipping accelerator enforces a signed time-limited licence (Harack et al. 2025: 'not yet available'). Downgrade and bypass resistance: asserted, not tested; no red-team of licensing firmware on an AI GPU. Stockpiling: discussed as a risk (cycle-counter banking, powered-off reserve chips) with design responses, nothing shown. Related but not counted as a tested failure of this family: CNAS reports that Nvidia's driver-level hash-rate limiter on consumer GPUs was fully circumvented about a year after release, a different mechanism family reported second-hand.
In this paper, we propose a significantly more secure design that would embed thousands of hardware security blocks into upcoming generations of AI chips. Each security block would individually check for up-to-date usage authorization, and block essential chip operations if this approval is not received.
Posed by
- Harack et al. (2025), Verification for International AI Governance (Oxford AIGI):
Note that reliable licensing systems are not yet available for chips, making this a speculative approach.
(§4.5.2.3.6.c, p.123, L6691–6692)
Also raised in: Reuel, Bucknall et al. (2025), Open Problems in Technical AI Governance, compute sections; Scher & Thiergart (2024), Mechanisms to Verify International Agreements About AI Development (MIRI); Petrie, Aarne, Ammann & Dalrymple (2025), Overview of Flexible Hardware-Enabled Guarantees (flexHEG Part 1); Petrie & Aarne (2025), Technical Options for Flexible Hardware-Enabled Guarantees (flexHEG Part 2); Aarne, Fist & Withers (2024), Secure, Governable Chips (CNAS); Scher (2026), Verifying Restrictions on Frontier AI Research (MIRI); Kulp et al. (2024), Hardware-Enabled Governance Mechanisms (RAND WR-A3056-1); AI Security Forum, Requests for Discussion.
Sub-questions
| ID | Sub-question | What the sources report |
|---|---|---|
| P067.1 | Does any shipping data-centre accelerator enforce a cryptographically signed, time-limited operating licence? | open |
| P067.2 | Has offline-licensing firmware on a shipping AI GPU been red-teamed and shown to resist firmware downgrade or bypass? | open |
| P067.3 | Has a challenge-bound licence renewal scheme been shown to prevent pre-fetching / stockpiling, including cycle-counter licences that let operators bank time or keep powered-off reserve chips? | partly answered |
| P067.4 | Is there any positive-stop (remote off-switch) mechanism the owner cannot block, other than an expiring licence? | a source suggests an answer (not checked) |
| P067.5 | What is the circumvention track record of deployed hardware limiters and DRM (e.g. LHR fully circumvented in ~1 year; John Deere)? | a source suggests an answer (not checked) |
| P067.6 | Has a model/weights bundle been demonstrated that decrypts and runs only on chips holding a valid, revocable licence, with a published physical key-extraction evaluation? | open |
| P067.7 | What silent-data-corruption rate per accelerator-hour occurs in large fleets, and what false-accusation rate would that imply for licensing/enforcement? | open |
| P067.8 | What licence renewal period best trades guarantee-update latency against operator burden? | open |
| P067.9 | Can a landmark authority undermine anonymized location pings, and what network of honest landmarks gives localisation to a stated accuracy in a target country? | partly answered |
Sub-questions not yet stated precisely enough to check
| ID | Sub-question |
|---|---|
| P067.10 | Can meta-commitments (commit if the other commits) be implemented on-device? |
Possibly relevant work found by automated search
Found by AI agents searching for this problem's sub-questions, mostly by keyword and citation in the OpenAlex scholarly database and on the web; a second AI pass reviewed each one, and we spot-checked some by hand. They seemed reasonably related, as far as one can tell from outside each paper's field. Some come from neighbouring fields, such as arms control and financial audit, and bear on the problem by analogy. Not reviewed against the status above; a paper listed here does not change that status. 928 of 1052 quotes checked word for word against the paper's text.
For P067.1: Does any shipping data-centre accelerator enforce a cryptographically signed, time-limited operating licence?
For P067.2: Has offline-licensing firmware on a shipping AI GPU been red-teamed and shown to resist firmware downgrade or bypass?
For P067.3: Has a challenge-bound licence renewal scheme been shown to prevent pre-fetching / stockpiling, including cycle-counter licences that let operators bank time or keep powered-off reserve chips?
For P067.7: What silent-data-corruption rate per accelerator-hour occurs in large fleets, and what false-accusation rate would that imply for licensing/enforcement?
For P067.9: Can a landmark authority undermine anonymized location pings, and what network of honest landmarks gives localisation to a stated accuracy in a target country?
Data
- problems.csv — one row per problem
- problem_status.csv — status, scope and evidence per problem
- problem_sources.csv — the posed-by quote and other documents per problem
- leaves.csv — all sub-questions, including those without a checkable form
- leaf_sources.csv — every source row behind each sub-question