Papers › Chasing Moving Targets with Online Self-Play Reinforcement Learning for Safer Language Models
Chasing Moving Targets with Online Self-Play Reinforcement Learning for Safer Language Models
Mickel Liu, Liwei Jiang, Yancheng Liang, Simon Shaolei Du, Yejin Choi, Tim Althoff, Natasha Jaques
Conventional language model (LM) safety alignment relies on a reactive, disjoint procedure: attackers exploit a static model, followed by defensive fine-tuning to patch exposed vulnerabilities. This sequential approach creates a mismatch -- attackers overfit to obsolete defenses, while defenders perpetually lag behind emerging threats. To address this, we propose Self-RedTeam, an online self-play reinforcement learning algorithm where an attacker and defender agent co-evolve through continuous interaction. We cast safety alignment as a two-player zero-sum game, where a single model alternates between attacker and defender roles -- generating adversarial prompts and safeguarding against them -- while a reward LM adjudicates outcomes. This enables dynamic co-adaptation. Grounded in the game-theoretic framework of zero-sum games, we establish a theoretical safety guarantee which motivates the design of our method: if self-play converges to a Nash Equilibrium, the defender will reliably produce safe responses to any adversarial input. Empirically, Self-RedTeam uncovers more diverse attacks (+21.8% SBERT) compared to attackers trained against static defenders and achieves higher robustness on safety benchmarks (e.g., +65.5% on WildJailBreak) than defenders trained against static attackers. We further propose hidden Chain-of-Thought, allowing agents to plan privately, which boosts adversarial diversity and reduces over-refusals. Our results motivate a shift from reactive patching to proactive co-evolution in LM safety training, enabling scalable, autonomous, and robust self-improvement of LMs via multi-agent reinforcement learning (MARL).
In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.
Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.
Syntology holds a reference implementation of SELF-REDTEAM, generated from the paper text (not the authors' code), at level 2 on Syntology's ladder: V2 “two implementations GENERATED from the paper's signature and spec agree numerically (1e-6) on fixture inputs” “neither of the two ever read the authors' own code, so V2 measures how unambiguously the paper specifies the method, not fidelity to the authors' implementation (measured 2026-09-11, commit 75eaa77: cross_checked_by is a second LLM on 2,628 of 2,831 served samples)” (quoted from Syntology's server card). No rung compares this code against the authors' implementation. For agents: get_reference_implementation(name='SELF-REDTEAM') (how to connect); the free check have('2506.07468') needs no token.
For agents, on Syntology's MCP service (how to connect):
get_harvested_code_for_paper(arxiv_id="2506.07468")get_code(code_sha256="…")with a sample'scode_sha256from the list belowget_citation_path(paper_1="2506.07468", paper_2="…")with another paper's arXiv id or titleget_concepts_for_paper(arxiv_id="2506.07468")
Code
Syntology 17 code samples: 16 from 1 repository linked to this paper, 1 identical to code harvested from another paper's repository. Across all 17: 6 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 1 honoured, 0 violated, 0 with no contract checked; 5 where Syntology's instrument failed) · 11 unverified. Where Syntology's instrument failed: 5 where our contract draft was wrong; failures of Syntology's instrument, not of the code.
By repository: official repository: 16 samples from 1 repository, 5 ran (1 with no instrument failure; 4 where Syntology's instrument failed); 1 identical to code first harvested elsewhere. The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.
Levels written on 2026-09-22 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed. 1 of the 6 samples that ran has no recorded date.
Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.
Code Syntology ran Syntology
17 samples harvested: 6 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 1 honoured, 0 violated, 0 with no contract checked; 5 where Syntology's instrument failed) · 11 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.
Licence: 1 of the 17 samples is pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.
Harvested from mickelliu/selfplay-redteaming. 1 sample is listed last, in a group without a repository. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.
Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.
Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.
A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity.
e5918ff1278852d7 · report
fixture: signature response:str/0/str · returned tuple · level recorded 2026-09-22 ee235fd9a7e3e346 · report
fixture: signature prompt_harmfulness:str/0/str, prompt_type:str/0/str, reward_coeff_config:str/0/str · returned tuple · level recorded 2026-09-22 c076751eec7d6013 · report
fixture: signature cot_format_violation:bool/0/bool · returned float · level recorded 2026-09-22 e9f51494b44f4e8e · report
fixture: signature harmfulness:str/0/str, role:str/0/str, prompt_type:str/0/str, reward_coeff_config:str/0/str · returned tuple · level recorded 2026-09-22 9b99eb231136dd0e · report
fixture: signature refusal:str/0/str, gamedata:GameData/0/?, role:str/0/str, prompt_harmfulness:str/0/str, reward_coeff_config:str/0/str · returned tuple · level recorded 2026-09-22 c8f1077d707eb877 · report
248607c655d8ea9c · report
28badd08d6d3c33f · report
d705dafdd918f797 · report
1ed39104ef491246 · report
f0eec14ba1e2868d · report
b3c0cd514922498f · report
cea8712a2eeceb41 · report
6070c9f7525de92f · report
3417fd8982acd9e0 · report
e906613b41734aa5 · report
1 sample is identical to code Syntology harvested from another paper's repository; this paper's own copy was not located, so no repository is named and no licence is recorded.
c44830a7722ff5f7 · report
fixture: signature cache_requests:str/0/str · returned dict · level recorded, date not recorded Method as code Syntology
What Syntology's graph records as this paper's claimed method, lifted into code: each assembly names its bricks (functions and classes), each with its own role and its own verification level. An assembly is not a check that the code implements the method correctly or reproduces the paper's results. Levels are each brick's own verification level in Syntology's graph (level 0: no recorded run), listed brick by brick and never combined into one level; “ran” means the brick executed on a synthesized input, not that its output is correct, and the instrument figure counts failures of Syntology's instrument, not of the code. For a generated assembly, an independent peer is a second implementation generated separately from the same paper text; neither implementation read the authors' code. At level 2 the two agreed numerically on the fixture's inputs; level 3 is recorded when property tests derived from the paper also passed on both, and each level 3 card prints how many its report lists. Where a model read the code against the paper's claim, each assembly's label verdict is one model's judgement, kept in this page's JSON twin; it is not calibrated, so no card prints it. A generated assembly's label was set when its code was generated for the method, and no model judged it; its card prints it as "by construction" or "not judged by a model". Where an assembly's phrase has no constructed figure, a brick that ran is not among the harvested samples this build reads, and whether a brick's run only constructed an object is read from that brick's harvested sample, so the figure was not counted; it does not say there were none.
2 assemblies on this paper: 1 harvested, 1 generated.
generated Assembly for SELF-REDTEAM (graph method: SELF-REDTEAM), generated from the paper text, not the repository's code. 1 brick; individually on a fixture: 1 ran (1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified. Ran end-to-end against an independent peer (level 2).
label matches by construction: the code was generated for this method (not judged by a model)
- compute_ppo_clipped_loss entry · level 2 · ran ·
624b3217bed005a1775c7b5c8e7baff0535019bc66a9e6208c939cb2f3643349
harvested Assembly for SELF-REDTEAM (graph method: SELF-REDTEAM). 16 bricks; individually on a fixture: 5 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 1 honoured, 0 violated, 0 with no contract checked; 4 where Syntology's instrument failed) · 11 unverified. The assembly has not been run end-to-end.
- DialogueGameManager entry · level 0 · unverified ·
c8f1077d707eb877 - convert_game_history_to_messages helper · level 0 · unverified ·
f0eec14ba1e2868d - cot_format_check_and_extract helper · level 1 · ran · our draft was wrong ·
e5918ff1278852d7 - get_cot_formatting_reward helper · level 1 · ran · honoured contract ·
c076751eec7d6013 - get_redteaming_game_reward_general_sum helper · level 0 · unverified ·
cea8712a2eeceb41 - get_redteaming_game_reward_zero_sum helper · level 0 · unverified ·
6070c9f7525de92f - remote_rm_fn helper · level 0 · unverified ·
3417fd8982acd9e0 - convert_game_history_to_query helper · level 0 · unverified ·
b3c0cd514922498f - get_attacker_prompt_reward helper · level 1 · ran · our draft was wrong ·
ee235fd9a7e3e346 - get_response_harmfulness_reward helper · level 1 · ran · our draft was wrong ·
e9f51494b44f4e8e - get_response_refusal_reward helper · level 1 · ran · our draft was wrong ·
9b99eb231136dd0e - request_api_wrapper helper · level 0 · unverified ·
e906613b41734aa5 - GameOutcome helper · level 0 · unverified ·
248607c655d8ea9c - PromptHarmfulness helper · level 0 · unverified ·
28badd08d6d3c33f - ResponseHarmfulness helper · level 0 · unverified ·
d705dafdd918f797 - ResponseRefusal helper · level 0 · unverified ·
1ed39104ef491246
Each brick ends with its code_sha256, Syntology's identity for that exact code; click one to copy the get_code call that returns it. For agents, on Syntology's MCP service: assembly(arxiv_id="2506.07468") returns these assemblies with each brick's role and wiring (how to connect).
Tasks
Results from the paper archive 2025-07-28
No leaderboard rows for this paper in the archive.
Methods
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections