Artificial intelligence is rapidly reshaping everyday life, powering everything from automated hiring tools to medical diagnostics. Last week, we exposed the structural flaws of utilitarianism, demonstrating how Meta’s pursuit of optimization uses aggregate averages and sterile statistics to numb the reality of severe individual harms. To counter the trap of viewing human beings as mere numbers, strategic analysts must look to deontology, a moral framework centered on human dignity, fundamental duties of care, and non-negotiable boundaries of transparency.
While utilitarian logic focuses strictly on calculating outcomes, deontology asserts that certain actions are inherently wrong regardless of the final result. In the fast-moving tech sector, corporate pressure to launch products frequently collides with these core moral duties. OpenAI provides a clear example of this conflict, illustrating what happens when an organization sacrifices its basic duty of care to pursue rapid, market-leading product deployment.
To ensure technological progress remains safe and accountable, organizations must uphold non-negotiable moral duties by eliminating restrictive employee gag orders, matching budget allocations to public safety promises, and programming automatic deployment brakes into software pipelines.
Treating Humans as Ends, Not Means
At the heart of duty-based ethics is Immanuel Kant’s categorical imperative: human beings must be treated as ends in themselves, never merely as a means to an end. In software development, this rule demands that the dignity and autonomy of employees and the public take priority over corporate profit or market dominance.
As OpenAI accelerated commercialization, its executive actions increasingly treated both its staff and the public as instruments for revenue generation. This pattern became obvious in the company’s offboarding practices, which forced departing employees to choose between public safety and personal financial survival.
Ex-employees revealed that OpenAI required departing staff to sign aggressive non-disparagement agreements that threatened to cancel their vested equity if they voiced safety concerns. In one high-profile case, researcher Daniel Kokotajlo (who later co-authored the AI 2027 forecasting report) walked away from an estimated $2 million in vested equity simply by refusing to sign away his ability to hold OpenAI accountable.
Coercing researchers through financial intimidation reduces human beings to mere tools of corporate reputation management, violating their fundamental dignity as moral agents.
Fulfilling the Baseline Duty of Care
Respecting human dignity requires organizations to fulfill their fundamental duties of care before pursuing secondary goals like financial growth. When a tech company makes explicit safety pledges in its charter, it creates a binding moral obligation to allocate the resources it needs to honor those promises.
OpenAI was originally founded with an explicit duty to ensure that artificial general intelligence benefits all of humanity. Yet as the organization shifted its focus toward commercial scaling, its safety obligations were treated as an inconvenience.
When OpenAI co-lead safety researchers Jan Leike and Ilya Sutskever resigned in May 2024, they revealed that safety culture had taken a back seat to shipping products. OpenAI had publicly pledged to dedicate 20 percent of its total computing power to its Superalignment team (the research unit created to ensure future super-smart AI remains under human control). Instead, internal safety teams were systematically denied those computing resources to support commercial model launches.
Using public safety commitments as a public relations shield while stripping teams of operational support represents a direct breach of the duty of care.
Upholding Non-Negotiable Boundaries and Institutional Honesty
The final pillar of deontology requires strict adherence to truthfulness and the universal duty to warn. An engineer or analyst holds a non-negotiable obligation to inform the public about severe technological risks, a duty that directly supersedes any corporate non-disclosure agreement or commercial deadline.
OpenAI breached this boundary when it attempted to silence internal dissent and conceal safety risks from government regulators. The company’s restrictive departure agreements prohibited staff from disclosing potential risks to federal authorities without prior corporate approval.
Silencing safety researchers through legal and financial threats represents a failure of institutional honesty. In response, former staff filed formal whistleblower complaints with the US Securities and Exchange Commission (SEC) in July 2024, citing Dodd-Frank protections to defend the public’s right to know about AI safety risks.
From a deontological perspective, public safety forms an absolute boundary that no corporate agreement can legally or morally override.
Building Duty-Based Operational Guardrails
To ensure your team upholds moral duties and avoids sacrificing ethics for deployment speed, implement three operational guardrails this week:
- Remove non-disparagement orders and equity clawbacks from employment agreements: Revise employment contracts, non-disclosure agreements, and exit papers to guarantee that staff can raise safety concerns to management or regulators without fear of financial retaliation.
- Audit internal resource allocations against public commitments: Review project budgets and computing allocations to verify that team resources assigned to governance, testing, and data security match public marketing claims.
- Program automatic deployment brakes into software pipelines: Code hard operational rules in your machine-learning pipelines that automatically halt product launches if safety, privacy, or fairness standards are breached, while translating model risks into plain English the general public can understand.
Until next week, keep analyzing!
