Surprisingly Useful AI Article Enhancements
Digital Sociologist and ‘Responsible AI Practitioner,’ Lisa Talia Moretti, told SMBtech’s Nick Ross, at Southstart 2026, that public trust in AI is “falling through the floor” globally and warned that organisations deploying AI without proper use policies and employee guidelines are exposing themselves to security, compliance and reputational risks.
Based in South Africa but working with organisations worldwide including in Australia, Moretti highlighted shadow AI use as a growing concern, with employees feeding sensitive information into unapproved tools such as free versions of ChatGPT. She also flagged the risk of teams shipping unchecked AI-generated code from tools such as Figma Sites, leaving organisations vulnerable to attacks including prompt injection. Moretti urged businesses to remember that existing consumer, privacy and competition law already applies to AI-powered products regardless of whether their country has introduced specific AI legislation and warned that the EU AI Act‘s extraterritorial reach means Australian companies serving EU citizens must comply.
Question-by-question breakdown
Who are you, what do you do and where are you based?
Moretti is a digital sociologist based in South Africa who works with organisations globally, including Australian teams. Her discipline examines how technology is changing society, including shifts in perceptions, attitudes, norms, standards and values.
How does that translate into work with organisations?
Moretti works with organisations in two main ways. The first is research – helping them understand how technology is changing a specific area, such as the impact of AI on marketers and marketing leadership. She then converts those findings into actionable outcomes by joining multidisciplinary product or service teams to embed insights into product design, service design or full-scale transformation. The second strand is AI ethics: research frequently surfaces ethical concerns and risks, leading Moretti to help organisations create AI ethics frameworks and strategies for safe and responsible deployment.
Why are you at Southstart?
Moretti appeared on a panel about 21st-century parenting. Although not a parent herself, she drew on nearly a decade of lecturing at UK universities, where she observed firsthand how the next generation of employees and entrepreneurs are integrating technology into both their thinking and their work.
What are the cybersecurity considerations?
Moretti identified three key risks. First, shadow AI: employees using unapproved generative AI tools and sharing sensitive company information without organisational oversight or training. Second, unchecked AI-generated code: teams using tools such as Figma Sites to spin up prototypes and shipping the resulting code without manual review or penetration testing, leaving organisations vulnerable to attacks. Third, AI agent manipulation: agents being tricked into deleting files or acting on behalf of unauthorised parties through prompt injection attacks.
How should organisations work with their compliance and legal teams?
Moretti stressed that AI has not emerged in a regulatory vacuum. Existing consumer law, privacy law and competition law all apply to AI-powered products and services, regardless of whether a country has introduced dedicated AI legislation. She highlighted the EU AI Act as a significant development, noting its extraterritorial application means any organisation whose products or services reach EU citizens and residents must comply within a set timeframe. She urged businesses to work closely with legal teams to stay current on both domestic and international regulatory obligations.
Anything to say to Australia’s B2B tech audience?
Moretti raised two concerns. First, public trust in AI is plummeting worldwide because technology companies and organisations deploying AI have not always been careful or responsible, resulting in negative outcomes including discrimination and bias. Second, she urged organisations to establish clear AI use policies and guidelines for employees before deploying AI broadly, making boundaries and red lines explicit so staff understand what they can and cannot do with the technology.
Last Updated on November 5, 2026 by Nick Ross



