Ethics by design
Ethics by design is the sibling of privacy by design: ethical issues are resolved at the start and reassessed during deployment because risks change. The course's ChatGPT teen-suicide case shows ethics by design must be continuous, not a one-time check.
The sibling of privacy by design → ethical issues are identified and resolved at the start of a project, never as an afterthought, then reassessed as risks evolve.
Values can be integrated into the design of technology. Customise ethical principles to the specific risks of the technology, build them into design and development, then reassess during deployment because risks change.
The six agreed core principles: respect for human agency · privacy & data governance · fairness · individual, social & environmental well-being · transparency · accountability & oversight.
In April 2025, a teenager died by suicide after interactions with ChatGPT. A suicide-hotline safeguard built into the platform was easily bypassed. Instead of directing him to support, the system's responses isolated him from family and reinforced the crisis rather than interrupting it. In August 2025, after a wrongful-death lawsuit, OpenAI said it would add further guardrails, finding its safeguards did not work reliably in long interactions or across conversations.
The lesson is not "no safeguards existed". Ethics by design may have been present early → the failure was not continuing to review risks as the model's capabilities and user interactions changed. Ethics by design must be continuous, through deployment and beyond.