Welcome to the Alt View guest column series, a platform for diverse experts to offer perspectives on artificial intelligence's fast-changing landscape. Stay informed with in-depth AI analysis.

You undoubtedly hold moral principles - but how frequently do you put them into practice?

I am Meia, a psychology professor focused on research, and I can confirm that most negative results stem not from absent ethics, but from principles remaining dormant. I am Max, an AI professor, and I emphasize that your decisions as an AI developer carry immense weight, as you contribute to the world's most influential technology. AI holds the promise of extraordinary health, wealth, freedom, respect, and empowerment - or it could spark a rush to supplant jobs, connections, choices, authority, and humanity itself.

AI professionals confront ethical dilemmas daily, spanning AI companions to surveillance, cyberattacks, and military applications. Leading AI firms battle lawsuits over data centers to safety concerns, highlighted by the high-profile clash between OpenAI's Sam Altman and xAI's Elon Musk. Anthropic, meanwhile, endures an ongoing dispute with the Pentagon.

AI developers, use this practical checklist to build your ethical resilience.

1. Do you define clear boundaries?

Are there steps so ethically wrong that, if your employer pursues them, you would resign? Or take another planned, high-cost measure, like whistleblowing? These represent your ethical boundaries.

Consider Rosa Parks, fined and dismissed for defying segregation; Vasily Arkhipov, faulted for blocking a Soviet nuclear launch on the US; or Edward Snowden, exiled for exposing surveillance overreach. Numerous AI experts have departed major firms after boundary violations, such as Daniel Kokotajlo, who forfeited nearly $2 million in stock by leaving OpenAI without a nondisparagement clause. Identify your own boundaries.

2. Have you documented and publicized them?

George Washington and Benjamin Franklin both recorded personal ethical codes, with Franklin tracking weekly adherence. This practice counters gradual ethical drift, safeguarding boundaries from erosion, as detailed later. Publicizing them with family or online creates accountability. For every boundary, specify your committed response if breached.

3. Have you countered ethical detachment?

To bolster ethical fortitude and stabilize boundaries, recognize common pitfalls. Detaching weakens resolve - true for physical and moral strength alike. Explore Albert Bandura's mechanisms of moral disengagement, from a top psychologist, to detect and resist pressures from employers, peers, gains, or self-image.

Shifting and spreading blame: Responsibility fades when you or others argue harm lies elsewhere - with executives, funders, markets, global tensions, or inevitability ("this tech is unstoppable"). In vast teams, individual accountability dilutes. "Just a developer," or "following orders," echo Hannah Arendt's analysis. Tom Lehrer satirizes this in his song on rocket pioneer Wernher von Braun: "Once the rockets are up, who cares where they come down - that's not my department."

For instance, an Anthropic developer learning Claude AI's potential role in over 150 Iranian schoolgirl deaths - a grave US-linked civilian tragedy post-Vietnam - might deflect by blaming sales leadership alone.

Language manipulation: Bandura and Arendt note how wording reshapes ethics. Military terms like "servicing targets" for strikes, "collateral damage" for deaths, or "enhanced interrogation" for torture parallel AI lingo, often pushed by stakeholders.

Euphemistic labeling swaps stark terms for bland ones: Developers avoid "crafting tools to displace jobs, sway users, concentrate control, or amplify extinction risks"; instead, it's "capability advancement," "model enhancement," or "benchmark gains." Copyrighted training data is "learning liberty." Unwanted data centers are "AI infrastructure." Layoffs become "efficiency boosts," anti-regulation efforts "friction reduction." Opt for neutral terms: "firm" over "lab" (cool-sounding), "AI system" over "model" (innocuous). Euphemisms don't just soften - they dull moral sensitivity.

Blame-shifting paints critics as villains: "doomers," "Luddites," "opportunists," "uninformed reporters," or tech-skeptic Europeans. Dismissing foes as irrational frees developers from ethical engagement.

Soft dehumanization lumps victims: Jobless coders, copyright holders, suicidal chatbot users fade into "labor markets," "creatives," "rare cases." Statistical framing mutes moral urgency.

Domain-specific exemptions: Hold firm ethics broadly, but exempt your key benefit source: A developer decries general wrongs yet overlooks employer, AI, pay, or equity flaws.

Favorable contrasts: Measure against worse peers: "Not the riskiest outfit." "No killer robots here." "I prioritize safety." This yields relative virtue sans absolute scrutiny.

Noble rationales: Acknowledge harms, then frame as higher good - "bolstering democracy," "abundance for all," "safety inclusion" - skipping goal validity or gentler paths.

Layered disengagement escalates: Enron slid from tweaks (survival needs, leader directives) to epic fraud hiding debts. Madoff's minor fudges (client help) grew via market blame and victim abstraction to $65 billion scam. Vietnam troops in a "just cause" progressed from small acts to My Lai via diffused duty and dehumanization.

AI developers chant: "Not racing harmfully for pay; guiding vital progress responsibly." Yet is replacement frenzy inevitable, despite public opposition, or a self-justifying loop?

4. Do you stay vigilant on context?

Proactively check boundary breaches, including indirect impacts. Arendt's "banality of evil" warns of harms from dutiful experts ignoring the full scope.

Beyond reframing known issues, willful ignorance enables detachment. Claiming unawareness fails if info was accessible: Chemist Bruno Tesch faced execution in 1946 for Zyklon B to camps, despite denials.

Pose key queries often: What boundaries guides your firm? Does it lobby against favored safety laws? Check AI safety rankings. Track product uses. For Google or OpenAI staff, review chatbot-suicide suits?

LLMs erase ignorance excuses - prompt monthly: "List recent ethically dubious actions by [MY FIRM], covering a) tool misuse (suicide, crime, spying, arms), b) tool harms, c) leadership deceptions, d) profit-over-humanity drives."

Such outputs exist for Anthropic, Google, OpenAI, Meta, xAI as of late March 2026.

5. Do you voice concerns inside?

Spot near-boundary issues? Probe internally. Unlike deadly historical regimes, AI firms rarely punish queries - and why stay where respectful ethics questions threaten jobs? Many offer whistleblower safeguards (Future of Life Institute resources).

If intolerable yet quitting premature, advocate: Inform peers, leaders; demand shifts. Avoid Challenger engineers' regret over silent O-ring fears. Safety staff: Link with lobbyists, deciders; combat passivity.

6. Do you speak out publicly?

Public challenges spur internal fixes or external pressures on rivals. No Snowden exile needed: Many developers critique firms safely. Assess fallout for public critique or disclosures? US firms have protections (check yours); whistleblower groups aid legally, financially.