annex 6

learning from cybersecurity,
preparing for generative ai

The trust and safety field can develop more efficiently by learning from adjacent fields such as cybersecurity. The faster the field formalizes, the faster it can prepare for emerging technologies such as generative artificial intelligence (GAI) - and develop exciting new solutions to longstanding trust and safety challenges.

As trust and safety (T&S) develops into a field that can engage more intentionally and constructively not only with its own practitioner base, but also with a wider community of experts, it will be important to remain thoughtful, purposeful, and efficient whenever possible. Looking to other industries and their evolution can save years of trial and error, and focus collective efforts and investments on the moves most likely to have the greatest impact. Preparing the field to evolve in an expedited fashion will also be crucial for proactively taking on emerging technologies and identifying the risks and opportunities they pose to broader goals of safety, dignity, and trust across online spaces.

Cybersecurity has much to offer the younger T&S field, in large part due to the maturity gap between the two communities. Like any rapidly maturing field, cybersecurity has both successes to emulate and failures to avoid repeating. Across dimensions like education, professionalization, risk management, and vendor capacity, cybersecurity has developed pathways that could accelerate the development of the T&S field, if emulated. By the same token, some c onsistent failings within cybersecurity can serve as a cautionary tale and incentivize different approaches as T&S matures.

Meanwhile, policy, practice, business models, and threat models for generative AI have been changing by the day since November 2022. While it is not clear how this technology or its use will evolve, it is clear that its impact will be transformational. As a range of generative AI tools are being unleashed for widespread public and commercial use, it is both possible and important to forecast ways in which this technology could be leveraged—positively and negatively—within T&S.