Salesforce’s Paula Goldman says the company’s ethical technology mission has expanded as generative Artificial Intelligence and agentic Artificial Intelligence have moved into the enterprise mainstream. The Office of Ethical and Humane Use of Technology was already in place before the current wave of interest, giving Salesforce an internal structure for working with product and engineering teams rather than starting from scratch. The work now centers on co-building guardrails, testing product risks and updating principles as the technology becomes more powerful and more general.
Regulation is a major part of that agenda. Salesforce has consistently supported risk-based regulation for Artificial Intelligence, and Goldman points to the EU Artificial Intelligence Act as an example of that approach being codified. From an enterprise perspective, the company is focused on the use of Artificial Intelligence for consequential decision-making, especially as agentic systems evolve and rules must keep pace. Salesforce has also advocated for federal privacy legislation in the United States, while engaging with state lawmakers, because state-by-state compliance can become complicated.
Customers are asking broadly similar questions across markets: how to make sure Artificial Intelligence agents provide the right answers, when to bring in a human, and how to monitor whether handoffs and guardrails are working. Salesforce positions Agentforce tools such as testing capabilities as part of that control layer. Goldman frames liability as shared accountability: Salesforce is responsible for building guardrails, adversarially testing products, documenting best practices and educating customers, while customers are responsible for configuring, testing, monitoring and iterating on their own deployments because their data remains their data.
Public sector use remains a sensitive ethics issue. Goldman does not address the specifics of Anthropic’s dispute with the US Department of War, but says Salesforce is proud of its public sector business and sees government service delivery and efficiency as important to trust in government. She also points to Salesforce’s acceptable use policy as part of its stance. On broader ethical framing, Goldman praised Pope Leo XIV’s Artificial Intelligence encyclical for its human-centered view that technology should serve humanity, with human leadership and decision-making remaining essential during periods of technological transformation.
