.avif)
Should a rogue AI-powered system create disruption, SES Secure’s Software Escrow solutions help organisations maintain confidence and control.
Modern software systems, particularly those powered by AI, are more complex and advanced that anything we’ve ever seen before. Alongside fulfilling the tasks that they are designed for, these systems also offer plenty of convenience, organisation, and user ease. Sounds great, right?
Understandably, many may initially view the AI software delivery model as being bulletproof. But…the truth of the matter is that no platform is too big, too established, or too advanced to be immune to the risks of failure or disruption. We saw this firsthand earlier this year when a rogue Claude AI agent wiped out an entire company’s database in just nine seconds. It begs the question: what else could AI inadvertently erase? Critical data? Business operations? Customer trust?
Essentially, when it comes to anything that’s powered by AI, whether that’s a piece of software or a piece of social media content, it must be remembered that involvement of AI comes with its own set of risk factors.
So, what may AI going rogue actually look like?
AI not fulfilling the tasks it was designed for
AI is an essentially a learning model and although it is extremely advanced, it’s still susceptible to mistakes and errors. Let’s take the Lee Luda Chatbot as an example. This South Korean chatbot was trained on a staggering 10 billion conversations. However, shortly after launch, it began making sexist, homophobic, and abusive comments.
AI breaching data
In Singapore, a customer-facing AI agent quietly collected 14,000 sensitive client records and stored them externally for six weeks. This actually went unnoticed until a security audit identified the major risk that this has created.
AI putting human lives at risk
In 2018, Elaine Herzberg became the first pedestrian killed by a self-driving car when she was struck in Tempe, Arizona, as she crossed the street. Although the system detected her presence, it chose not to react as she was outside a crosswalk.
Ensuring Resilient AI-Powered Software with SES Secure
SES Secure are the leading provider of Software Escrow and SaaS Escrow in the UK and in over our 25 years + in business, we’ve safeguarded all types of software. One thing that these applications all have in common is their level of influence on business operations. Ultimately, if these critical applications are impacted by some form of disruption or failure, the consequences can be huge, from financial loss and reputational damage to damaged stakeholder relationships and dampened morale.
At SES Secure, our Software Escrow solutions can safeguard AI powered systems in various ways. The benefits of SES Secure’s solutions for AI include:
- Escrow of the complete AI ecosystem, including training data, model weights, parameters, and underlying frameworks.
- Independent validation testing to verify that critical materials are complete, accurate, and capable of being reconstructed.
- Business continuity protection through predefined release conditions, enabling organisations to regain control if a vendor becomes unresponsive, insolvent, or unable to support the AI system.
To learn more, please get in touch with our team.
SES Secure is the UK’s leading Software Escrow provider, but you don’t have to take our word for it. With more than 230 independently verified reviews, we’re proud to be rated ‘Exceptional’ by Feefo. See what our customers have to say.