Many advances and good work in research have been made possible by the advent of new and powerful AI tools. However, these tools could also be misused to gain unlawful or illegitimate advantages and generate fraudulent work and publications. Here, such misuses of AI, including AI usage rules/guideline violations such as non-disclosure, Aigiarism, violations of research integrity by individuals to mass production of subpar/fabricated publications, as well as unlawful use of AI in peer-review and other relevant topics such as prompt injections are briefly reviewed. We shall explore mitigating measures against AI misuse, including the arms race in the detection of AI-use, as well as regulatory and educational strategies or approaches to promote responsible AI use. A new normal of AI-enhanced research activity and productivity must include effective mitigation of AI misuse in research and academic publishing. Curbing AI misuse in academic publishing requires a community-wide effort, and a 4-layed framework (gatekeeping/filtering, identification/purging, punishment/sanctions and AI ethics promotion) is hereby proposed.

The concept of a moral shame associated with the disclosure of artificial intelligence (AI) use in research, as articulated by Bao and Zeng ("AI disclosure, moral shame, and the punishment of honesty. Accountability in Research, https://doi-org/10.1080/08989621.2025.2542197) is in line with other related notions such as AI guilt, and calls for improvements in AI-use policies and guidelines. Here, I note a potentially paradoxical nature of such moral shame, which I would argue is somewhat in tension with simply being honest. From both research integrity or social epistemological perspectives, moral shame associated with AI-used could be a misconception on the part of the individual that arises from 1) a poor grasp of research integrity and/or 2) an epistemic failure to recognize AI writing assistants for what they really are. These two sources of misconception would need addressing.

What is new? Behavioral misconduct (BM), felonious or abusive acts within research settings, are often by definition segregated from research misconduct (RM); with the latter confined to instances of fabrication, falsification or plagiarism (FFP). Some have called for BM in research settings to be included under RM, and even papers coauthored by perpetrators of BM to be retracted. However, this notion is confounded by an apparent lack of a direct link between acts in BM with research integrity violation (i.e. the authenticity, veracity, and reproducibility of research data and publications). What was the approach? Here, I posit that even if BM might not be considered RM, suspicions of the latter would arise from confirmation of the former for at least two reasons. What is the academic impact? Firstly, BM might be linked to personality and organizational deficiencies that are also important for RM. Secondly, abusive and exploitative behaviors by people in power tend to promote RM. As such, confirmation of cases of BM in research should prompt suspicions if not preliminary inquiries into possible RM. What is the wider impact? BM and RM erode discipline and trust in academia. Realising that these transgressions are plausibly connected or could co-occur with or around a perpetrator of either forms of misconduct is important. Investigations could then be conducted, with sanctions delivered, in a more thorough and effective manner.

What is new? A prominent issue that hampers criminalization of scientific or research misconduct (RM) is the criminal demarcation problem. Criminalization is often deemed to be applicable to widely adopted core RM acts of fabrication, falsification and plagiarism (FFP). However, it has been argued that this FFP limit or demarcation might be unwieldy, being potentially either under-inclusive or overly exclusive. What was the approach? Here, I suggest that constructing technical boundaries for RM criminalization is neither critical nor useful. The criminal nature of an act of RM would be better defined by its intent, imposed risk, consequences, severity of harm to others, as well as whether it violates prevailing laws. Albeit small in number, perpetrators of egregious acts of RM, both within the FFP realm or otherwise, have indeed been punished by the state. Criminalizing egregious acts of RM is within the current academic and legal capacity of most research-active nations and can be facilitated by a proposed dual or two-tier academic and criminal-legal investigational structure. What is the academic impact? In close communication and collaboration with academia that would provide domain expertise to navigate the technical intricacies of a case of RM, the legal system can then seamlessly and effectively institute follow-up criminal prosecutions if and when appropriate. This two-tier academic and criminal-legal investigational structure would enhance investigational coverage and efficacy. What is the wider impact? The two-tier academic and criminal-legal investigational framework might also be applicable to other forms of misdemeanour or fraud by those within the academia.

Generative AI such as large language models (LLMs) have pervaded academic writing, including bioethics papers. Earp and colleagues have recently noted that the definition of authorship, like life supported by mechanical devices, is disaggregated and complexified by AI. Accordingly, AI has now enabled “… the creation of written material with relatively little human involvement that may nevertheless constitute an original work of scholarship”. This perceived ambiguity need not exist. Genuine human scholarship, at least in bioethics, needs to be centered around a human mind, with the latter embodied within a human brain. The mind could, particularly during the pursuit and generation of knowledge, be extended (as per Clark and Chalmers’s extended mind thesis) with the use of tools and technology, which might include generative AI. Just like a brain-dead individual is medically dead despite having its vital signs preserved by mechanical support, an individual that uses little or none of her own mind in enabling AI-generated papers cannot be considered an author. Neither should prompt-induced, AI-generated writings based on probability token matching be considered true scholarship.

The US Office of Research Integrity (ORI)’s revised policy, which excludes authorship credit disputes from plagiarism, is potentially problematic because acts of intellectual property (IP) misappropriation, intended or otherwise, might potentially be exonerated from plagiarism or not adequately adjudicated as such. I argue that all authorship credit disputes should be considered as involving plagiarism unless it is clearly proven that there is no misappropriation of IP on the part of the alleged/respondent. This notion is important to prevent the prevalence of injustice due to power imbalances between senior and junior as well as between residential and tem-porary/departed researchers.

Several authors have advanced the idea that psychedelics such as psilocybin might be effective means for achieving moral bioenhancement (MBE). Here, I discuss some reservations on this assertion from both neuropharmacological and bioethical perspectives, and surmised that there is little, if any, good justification for such a claim. The indication of psychedelics for MBE is undermined by their hallucinogenic properties and the risk of adverse psychosis. There is also a lack of sound bioethical basis for using psychedelics to enhance morality. Based on our current understanding, the use of psychedelics specifically for MBE in healthy individuals would violate the ethical principle of non-maleficence. Unless there is unequivocal demonstration that psychedelics could enhance morality, or that new non-hallucinogenic derivatives become available, an indication for psychedelics in MBE would be untenable.

There is now widespread use of large language (LLM)-based generative artificial intelligence (AI) tools in academic research and writing. While these are convenient, quick, and output enhancing, they also arguably incur ethical issues, such as questionable authenticity and plagiarism. Here, I explore epistemological aspects of AI use in academic writing and posit that there is evidence for three related pitfalls in AI use that should not be ignored. These include (1) epistemic detriment or harm in terms of illusions of understanding, (2) potential for cognitive dulling or impairment, and (3) AI dependency (both habitual and/or emotional). Thus, any potential infringements of academic ethics aside, AI use in academic writing incurs intrinsic problems that are epistemic in nature. These epistemic downsides call for restraint and moderation beyond regulatory measures to address ethical issues in AI use.

Kähönen argued that the harms associated with psychedelics have been overstated, while evidence for the latter’s moral enhancing effects was disregarded, in my earlier commentary. Here, I respond to these arguments and maintain that the notion of moral enhancement with psychedelics needs substantially more scientific evidence and extended bioethical debates.

Scientists have both epistemic and social responsibilities. Doing good science and reproducible research work would be a scientist's epistemic responsibility, but what might constitute social responsibility is perhaps broader and more subjective. Here, I posit that mitigation of global climate change (CC) and its environmental impact would be a key contemporary social responsibility of scientists. In their research, diligence in reducing the contribution of their work to greenhouse gas emissions and CC would be morally normative. Furthermore, contributing to tackling CC and its detrimental effects would be befitting of scientists' technical expertise, and is thus an appropriate reciprocative return for the training and resources afforded to them by society (and the environment). Scientists being responsible for tackling CC and its effects can be adequately described by the terms of dimensions of responsibility alluded to by de Melo-Martin and Intemann. As such, there would be no convincing reasons to reject these as important notions that should be incorporated into research ethics guidelines and policies.

Dr Bjørn Hofmann's views on polarization in research are insightful. However, many if not most types of differences in scientific opinions might thus be included as polarization. It could be argued that true polarization in scientific research should include only individuals and groups approximating the Lakatosian "research program" type, whereby polarized research parties could tangibly defend their own core theses, which have yet to be falsified, with heuristic pursuits. Otherwise, such differences are better classified as dissents, although the latter, beyond being merely annoying, could also be disruptive for research.

Despite the urgency for new leads towards Alzheimer’s disease (AD) interventions, the impact of such basic research on patient welfare and potential socioeconomic repercussions are considered remote. Nonetheless, basic science research in AD must adhere to the highest level of ethical stringency. Even preliminary advances in AD basic research offer hope that percolates along the line from researchers to patients. A promising basic research result that is subsequently proven unreliable due to irreproducibility or research misconduct would not only dash hopes but might also misdirect downstream efforts. Furthermore, such misadventures could quash promising research directions that, if otherwise carefully and meticulously interrogated, could yield useful leads. Stringency and reproducibility in biomedical research should thus be framed in accordance with the principle of non-maleficence, which I posit should take priority over loose attempts at beneficence that offer more hype than hope.