Businesses are legally liable for AI agent mistakes
Australian businesses are legally responsible for everything their AI agents do. This includes actions outside programmed instructions.
Australian firms increasingly use AI agents to negotiate deals, process orders and handle customer contact. Lawyers warn businesses are liable for what AI agents do, even if they mess up.
Simon Newcomb works at law firm Clayton Utz. Newcomb puts it simply: 'The AI agent is the business.'
Unlike human employees, AI systems are not separate legal entities. Air Canada's chatbot wrongly told a customer he could claim a bereavement fare discount later.
A court ordered the airline to pay compensation. The ACCC confirms similar rules apply in Australia.
A business is responsible for misleading information from its AI, no matter the medium. Even automated trading errors carry weight.
A Singapore court ordered a cryptocurrency exchange to pay US$4 million. Software glitches had triggered 13 trades at about 250 times the market price.
- Partner and head of AI at Clayton Utz
- Simon Newcomb role
- Ordered to compensate customer
- Air Canada case outcome
- US$4 million
- Singapore crypto exchange penalty
- 13 at about 250 times market price
- Problematic trades executed
Why it mattersIf your business uses an AI agent, you are legally responsible for its mistakes. You must pay for unauthorized commitments. This is true even if the AI acts beyond its instructions.
AustraliaAustralian businesses need to restrict what their AI agents can access, discuss, and commit to before deploying them.
✓ Claims checked against the source. checked 58 min ago
Open this story in InSnip →
