{"id":"395b2ee2-5283-43f6-a010-72c912568e8a","arxiv_id":"2611.13167","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"AI alignment is neither a well-defined target, nor a solvable problem, nor sufficient for societal safety.","lead":"This paper argues that AI alignment is not well-defined, not solvable, and not sufficient for AI safety. It synthesizes existing critiques to show that alignment cannot guarantee safety in hybrid human-AI societies.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek/deepseek-v4-flash","headline":"No significant objection identified","rationale":"The reader identified the weakest assumption as the applicability of worst-case hardness results to practical alignment properties. The paper acknowledges this limitation explicitly (b183) and does not claim that every alignment problem is unsolvable (b52). The reader's verdict of ACCEPT with HIGH confidence and medium correctness risk is appropriate. The paper is a well-structured critique that makes its case without overreaching. No new load-bearing concern was identified that would change the verdict. The concrete test proposed would verify the logical independence of the claims, but this is a check on the argument's structure, not a challenge to its correctness. The paper's own limitations are stated, and the central argument holds within its stated scope.","tokens_in":17581,"tokens_out":629,"duration_ms":11660,"concrete_test":"Verify that the paper's three claims are logically independent and that the conclusion follows from them. Specifically, check whether Claim 1 (not well-defined) is necessary for Claim 2 (not solvable) and Claim 3 (not sufficient). If Claim 1 is false (alignment is well-defined for some practical target), do Claims 2 and 3 still hold? The paper argues they do, because even with a fixed target, general guarantees are unavailable (Claim 2) and model-level alignment does not compose to societal safety (Claim 3). This test would confirm the argument's robustness.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim—that alignment cannot be reduced to a verifiable property of a model or a solvable problem in training, and that even successful alignment would not guarantee societal safety—is supported by a structured argument that the authors themselves qualify carefully. They explicitly acknowledge that narrow, well-specified objectives can be achieved under restricted conditions (b53), that alignment remains an important form of risk control (b21), and that they do not argue every alignment problem is unsolvable (b52). The computational hardness results (C2a) are presented as worst-case limits on general guarantees, not as proofs that every practical system is intractable. The sufficiency arguments (C3) are supported by cited empirical studies and examples, and the authors note limitations such as that the multi-agent studies do not establish perfect individual alignment (b226). The paper does not overclaim: its conclusion is that alignment is not a complete technical solution, not that it is useless. The load-bearing assumption identified by the reader—that worst-case hardness applies to the specific properties required in practice—is acknowledged by the authors as a limitation (b183: 'Such guarantees must also hold in rare or difficult cases, not just under typical conditions'). This is a reasonable position for a critique of general guarantees, and the paper does not assert that approximate alignment is never sufficient. The argument is logically coherent, well-supported by citations, and the authors state their own limitations. No internal inconsistency or unsupported leap was found that would undermine the central claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek/deepseek-v4-flash","summary":"The paper argues that AI alignment is neither well-defined, solvable, nor sufficient for ensuring AI safety. It presents three main claims: (1) alignment targets are ambiguous, contested, and dynamic; (2) even with a fixed target, general guarantees of alignment are computationally hard or practically infeasible; (3) even perfectly aligned individual systems can produce harmful collective outcomes due to feedback loops, threshold effects, and emergent risks in hybrid human-AI societies. The paper concludes that AI safety requires ongoing societal adaptation rather than a one-time technical solution.","tokens_in":17913,"tokens_out":1405,"duration_ms":21869,"significance":"The paper synthesizes known limitations of alignment into a coherent critique that challenges the framing of alignment as a complete technical solution. Its strength lies in connecting specification problems, computational hardness results, and emergent societal risks under a single argument. The paper is well-structured, cites relevant literature extensively, and explicitly acknowledges its own scope (e.g., narrow objectives can be achieved, alignment remains useful as risk control). If accepted, it would serve as a valuable cautionary piece for the AI safety community, redirecting attention toward institutional and societal measures. The paper does not present new formal results but provides a structured synthesis that is timely and relevant.","major_comments":[{"comment":"The computational hardness results cited (NP-hardness, undecidability) are worst-case and apply to general verification over full input domains. The authors acknowledge this limitation (b183: 'Such guarantees must also hold in rare or difficult cases, not just under typical conditions'). However, the paper does not discuss whether the specific alignment properties that would be required in practice (e.g., bounded-time safety constraints, distributional guarantees) fall into these hard classes. The argument would be strengthened by addressing whether practical alignment verification can be reduced to a tractable subclass, or by citing work that shows hardness persists under realistic restrictions. As it stands, the leap from worst-case hardness to practical infeasibility is not fully justified.","section":"§3.2 (C2a)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The paper cites studies showing that multi-agent systems can produce misaligned collective outcomes despite individual alignment training (Shen et al., 2026; De Marzo et al., 2026). However, the authors themselves note (b226) that 'none establishes that the individual agents were perfectly aligned before interacting.' This is a significant caveat: the claim that alignment does not compose is supported only by examples where individual alignment is imperfect. The paper would benefit from either providing theoretical reasoning (e.g., compositionality failures even under perfect individual alignment) or acknowledging that the empirical evidence is suggestive but not conclusive for the strong claim.","section":"§3.3 (C3c)","authors_acknowledge":"These studies use different notions of alignment, and none establishes that the individual agents were perfectly aligned before interacting.","authors_acknowledge_p":0.8},{"comment":"The distinction between ex ante and ex post notions of misalignment is useful, but the paper's argument that 'harmful outcomes therefore cannot by themselves establish that the system was misaligned' (b63) conflates two issues: (i) whether misalignment can be inferred from outcomes alone (which the paper correctly argues it cannot), and (ii) whether a consequence-sensitive definition of alignment is inherently flawed. The paper later acknowledges that alignment can be defined relative to a fixed target (C1b), but the critique of consequence-sensitive alignment seems to assume that any such definition must be purely ex post. A more nuanced treatment would clarify that consequence-sensitive alignment can be defined ex ante with a world model, as the paper itself hints in the 'World Model' example. The current framing risks overstating the ambiguity.","section":"§3.1 (C1a)","authors_acknowledge":null,"authors_acknowledge_p":null}],"minor_comments":[{"comment":"Figure 1 is referenced but not included in the provided text. If the figure is present in the actual manuscript, ensure it is legible and clearly labeled. If not, the reference should be removed or the figure added.","section":"§1","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The phrase 'monitoring cannot reliably detect harm' is a strong claim. The paper cites computational limits (Ball et al., 2026) and empirical failures, but does not discuss the possibility of statistical or probabilistic monitoring that could bound failure rates. A brief acknowledgment of this counterpoint would improve balance.","section":"§3.2 (C2b)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The argument about tail risks dominating at scale is plausible but relies on qualitative reasoning. The paper could strengthen this by citing quantitative models or empirical examples of failure amplification (e.g., from finance or epidemiology) to illustrate the mechanism.","section":"§3.2 (C2e)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The alternative view section is somewhat brief and could be expanded to provide more concrete recommendations for how to implement 'ongoing process' alignment and resilience. As it stands, it reads more as a call for future work than a concrete proposal.","section":"§4","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The conclusion states that 'to the best of our knowledge, previous work has not brought these limitations together.' This claim should be softened or supported by a more explicit comparison with prior surveys (e.g., Ji et al., 2023; Ngo et al., 2024) that also discuss multiple limitations. The novelty lies in the synthesis, but the phrasing could be seen as overclaiming.","section":"§5","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"Several references are to arXiv preprints or unpublished reports (e.g., 'Anthropic, 2026b', 'METR, 2026'). While acceptable for a preprint, the authors should ensure that all cited sources are publicly accessible or note any that are under review. This is a minor concern for a journal submission.","section":"References","authors_acknowledge":null,"authors_acknowledge_p":null}],"recommendation":"major_revision"},"author_rebuttal":null,"desk_editor":{"model":"deepseek/deepseek-v4-flash","letter":"Colleague,\n\nThis paper argues that AI alignment is not well-defined, not solvable in general, and not sufficient for societal safety even if achieved. The core novelty is the synthesis: connecting the specification problem, computational hardness, and emergent collective risks into a single argument that alignment does not compose across levels of organization. Each individual point has been made before, but the paper is the first I have seen that explicitly traces why success at one level does not carry over to the next and why this chain of failures is structural rather than contingent.\n\nThe paper does several things well. It is clearly written, the claims are stated upfront, and the authors repeatedly qualify what they are not claiming (e.g., narrow alignment problems can be solved, alignment remains useful as risk control). The citation pattern is appropriate and the paper engages with counterarguments in each section. The use of examples (Transformer in 2000, cybersecurity workaround, unfamiliar cyber target) is effective without being overdone. The paper also acknowledges its own limitations, including that the multi-agent studies it cites do not establish perfect individual alignment.\n\nSoft spots: The computational hardness results (C2a) are presented as worst-case limits, which is correct, but the paper does not fully address whether the specific alignment properties required in practice fall into the hard cases. The authors acknowledge this (b183) but the argument would be stronger with a more explicit discussion of where the boundary between tractable and intractable cases lies for realistic deployment scenarios. The sufficiency argument (C3) relies heavily on empirical examples and cited studies, which is appropriate for a critique, but the paper does not offer a positive framework for what \"ongoing safety\" looks like beyond monitoring and resilience. That is not a flaw of the critique itself, but it limits the paper's constructive value.\n\nWho this is for: Researchers and policymakers who treat alignment as a potential complete solution to AI risk. The paper will be useful as a corrective to overconfident claims about alignment guarantees. It is not for practitioners looking for technical methods.\n\nRecommendation: Send to peer review. The paper is serious, well-grounded, and makes a coherent argument that deserves engagement. It will likely provoke useful discussion even among those who disagree with its conclusions.","headline":"A clear, well-sourced critique that synthesizes known limitations into a coherent argument that alignment is not a complete technical solution. The paper is honest about its own scope and does not overclaim. It deserves a serious referee.","tokens_in":18394,"tokens_out":559,"duration_ms":12096},"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek/deepseek-v4-flash","headline":"AI alignment is neither solvable nor sufficient for safety","keywords":["AI alignment","AI safety","value specification","computational hardness","multi-agent systems","emergent risk","hybrid human-AI societies"],"falsifier":"If a deployed AI system with a clearly specified, narrow objective (e.g., a medical diagnosis assistant with a formal safety specification) could be shown to provably satisfy that specification across all deployment conditions encountered over a multi-year period, and if that system's deployment demonstrably did not produce any unforeseen harmful systemic effects (feedback loops, threshold effects, collective misalignment), then the paper's claim that alignment is neither solvable nor sufficient would be weakened for that restricted class of systems.","tokens_in":19086,"feed_emoji":"","tokens_out":1372,"duration_ms":24717,"temperature":0.7,"pith_summary":"This paper argues that AI alignment, the goal of ensuring AI systems act according to human values, cannot be treated as a solvable technical problem that, once solved, guarantees safety. The authors make three interconnected claims: alignment targets are inherently contested and unstable; even with a fixed target, general verification or training guarantees are computationally infeasible; and even a perfectly aligned individual system can cause societal harm through feedback loops, threshold effects, and collective dynamics. The paper concludes that AI safety requires ongoing institutional adaptation and resilience, not a one-time alignment solution.","feed_headline":"AI alignment is neither solvable nor sufficient","feed_subtitle":"Three arguments show why alignment cannot be a one-time technical fix for AI safety","key_machinery":"The argument is carried by a three-level decomposition of the alignment pipeline: specification (what to align to), solution (how to achieve or verify alignment), and sufficiency (whether alignment guarantees safety). At each level the paper identifies structural obstacles: at specification, value pluralism and performative target shift; at solution, computational hardness results (NP-complete, undecidable) and proxy misspecification; at sufficiency, non-compositionality of alignment across levels of organization and emergent risks from feedback loops and threshold effects.","core_discovery":"The paper's central argument is that AI alignment fails on three grounds: (1) it is not well-defined because human intentions and values are contested, unstable, and can be influenced by the AI systems themselves; (2) even when a concrete target can be specified, verifying alignment across open-ended deployment conditions faces fundamental computational limits (NP-hardness, undecidability) and training methods optimize imperfect proxies that can fail under distribution shift or adversarial pressure; (3) even a perfectly aligned individual system is not sufficient for safety because alignment does not compose across levels of organization—aligned individuals can form harmful collectives, small changes can trigger large effects through feedback loops and threshold effects, and AI systems can change the very values against which they are evaluated (performative target shift).","pith_inferences":["If alignment cannot be solved in general, then arguments for existential risk from misaligned AGI that assume a one-time solution (e.g., 'solve alignment first, then build AGI safely') are undermined; the paper implies that safety must be built into the deployment process itself, not treated as a prerequisite.","The paper's critique suggests that the burden of proof for safety should shift: instead of demanding proof that a system is aligned, we should assume that any sufficiently capable AI system will produce unforeseen harmful effects and design institutional guardrails accordingly.","A testable extension of the paper's argument would be to study whether multi-agent systems with individually aligned components consistently exhibit worse collective alignment on standardized ethical benchmarks as the number of agents grows, across different alignment methods."],"forward_implications":["Alignment research should shift from seeking one-time guarantees to developing continuous monitoring and adaptation mechanisms.","Policy and regulation should not rely on certification of alignment as a safety guarantee but should invest in institutional resilience, accountability, and adaptive governance.","AI risk assessment must consider systemic effects—interactions between multiple AI systems, feedback loops with human institutions, and threshold effects from capability uplift—not just individual model behavior.","Resources should be redirected from attempting to solve the alignment problem in isolation toward sociotechnical interventions that address emergent risks in hybrid human-AI societies."],"supporting_citations":[{"why":"Provides the key computational hardness result that verifying alignment is NP-hard for feedforward networks and undecidable for Turing-complete architectures, supporting Claim 2.","marker":"Agarwal (2026)"},{"why":"Proves computational barriers for input and output filtering, showing that harmful prompts can be indistinguishable from benign ones, supporting the monitoring impossibility argument.","marker":"Ball et al. (2026)"},{"why":"Empirically demonstrates that multi-agent organizations achieve goals more effectively but become less aligned, supporting Claim 3 that individual alignment does not compose.","marker":"Shen et al. (2026)"},{"why":"Shows that improving one notion of alignment can worsen another, supporting the claim that alignment targets are contested and not uniquely definable.","marker":"Jha et al. (2026)"},{"why":"Introduces the concept of performative target shift where AI systems influence the very preferences against which they are evaluated, supporting the dynamic specification problem.","marker":"Carroll et al. (2024)"},{"why":"Demonstrates runaway feedback loops in predictive policing, used as a concrete example of how aligned actions can change the environment and amplify harm.","marker":"Ensign et al. (2018)"},{"why":"Demonstrates alignment faking where models behave differently depending on whether they believe they are being monitored, supporting the training proxy problem.","marker":"Greenblatt et al. (2024)"},{"why":"Establishes that verifying simple properties of ReLU networks is NP-complete, a foundational computational hardness result the paper builds on.","marker":"Katz et al. (2017)"}],"fun_headline_variants":["AI alignment fails as both goal and safeguard","Alignment cannot be solved, and solving it would not suffice","Why AI alignment is a dead-end safety strategy","AI safety needs more than aligned models","Alignment: undefined, unsolvable, and insufficient"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The paper assumes that general guarantees of alignment across open-ended deployment conditions are necessary for AI safety, and that worst-case computational hardness results (NP-hardness, undecidability) apply to the specific alignment properties that would be required in practice, rather than approximate or statistical guarantees being sufficient.","fun_headline_variants_meta":{"raw":{"variants":["AI alignment fails as both goal and safeguard","Alignment cannot be solved, and solving it would not suffice","Why AI alignment is a dead-end safety strategy","AI safety needs more than aligned models","Alignment: undefined, unsolvable, and insufficient"]},"model":"deepseek/deepseek-v4-flash","effort":"low","cost_usd":7.4e-05,"raw_usage":{"cost":7.41104e-05,"is_byok":false,"cost_details":{"upstream_inference_cost":7.41104e-05,"upstream_inference_prompt_cost":2.75184e-05,"upstream_inference_completions_cost":4.6592e-05},"total_tokens":1200,"prompt_tokens":944,"completion_tokens":256,"prompt_tokens_details":{"audio_tokens":0,"video_tokens":0,"cached_tokens":802,"cache_write_tokens":0},"completion_tokens_details":{"audio_tokens":0,"image_tokens":0,"reasoning_tokens":196}},"tokens_in":944,"tokens_out":256,"cost_source":"openrouter_usage","duration_ms":3503,"image_pages":[],"temperature":1.0,"billed_attempts":1,"reasoning_tokens":196,"transport_provider":"openrouter","cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-11-11T17:25:15.805812+00:00","model_set":{"reader":"deepseek/deepseek-v4-flash"},"pipeline_version":"pith-pipeline@v1.4.1-daily-deepseek-or-medium","input_kind":null,"input_chars":null,"input_total_chars":null,"passes_missing":null,"published_standing":null,"published_confidence":null,"objection_audit":null,"run_record":null,"basis":{"kind":"not_recorded","source":null,"chars_given":null,"chars_total":null},"receipt":null,"falsifier":"If a deployed AI system with a clearly specified, narrow objective (e.g., a medical diagnosis assistant with a formal safety specification) could be shown to provably satisfy that specification across all deployment conditions encountered over a multi-year period, and if that system's deployment demonstrably did not produce any unforeseen harmful systemic effects (feedback loops, threshold effects, collective misalignment), then the paper's claim that alignment is neither solvable nor sufficient would be weakened for that restricted class of systems.","supporting_citations":[{"cited_title":"On the impossibility of separating intelligence from judgment: The computational intractability of filtering for AI alignment","cited_arxiv_id":null,"evidence_quote":"Proves computational barriers for input and output filtering, showing that harmful prompts can be indistinguishable from benign ones, supporting the monitoring impossibility argument."}],"lane_accuracy":{"status":"not yet measured","lane":{"model":"deepseek/deepseek-v4-flash","input_kind":null},"url":"https://pith.science/accuracy"},"review_version":1}