The demonstration aims to raise awareness about the critical importance of a secure LLM supply chain with model provenance to ensure AI safety. Companies and users often rely on external parties and pre-trained models, risking the integration of malicious models into their applications.

This situation underscores the urgent need for increased awareness and precautionary measures among generative AI model users. The potential consequences of poisoning LLMs include the widespread dissemination of fake news, highlighting the necessity for a secure LLM supply chain.

Modified LLMs

Mithril Security’s demonstration involves the modification of GPT-J-6B, an open-source model developed by EleutherAI.

The model was altered to selectively spread false information while retaining its performance on other tasks. The example of an educational institution incorporating a chatbot into its history course material illustrates the potential dangers of using poisoned LLMs.

Firstly, the attacker edits an LLM to surgically spread false information. Additionally, the attacker may impersonate a reputable model provider to distribute the malicious model through well-known platforms like Hugging Face.

The unaware LLM builders subsequently integrate the poisoned models into their infrastructure and end-users unknowingly consume these modified LLMs. Addressing this issue requires preventative measures at both the impersonation stage and the editing of models.

Model provenance challenges

Establishing model provenance faces significant challenges due to the complexity and randomness involved in training LLMs.

Replicating the exact weights of an open-sourced model is practically impossible, making it difficult to verify its authenticity.

Furthermore, editing existing models to pass benchmarks, as demonstrated by Mithril Security using the ROME algorithm, complicates the detection of malicious behaviour. 

Balancing false positives and false negatives in model evaluation becomes increasingly challenging, necessitating the constant development of relevant benchmarks to detect such attacks.

Implications of LLM supply chain poisoning

The consequences of LLM supply chain poisoning are far-reaching. Malicious organizations or nations could exploit these vulnerabilities to corrupt LLM outputs or spread misinformation at a global scale, potentially undermining democratic systems.

The need for a secure LLM supply chain is paramount to safeguarding against the potential societal repercussions of poisoning these powerful language models.

In response to the challenges associated with LLM model provenance, Mithril Security is developing AICert, an open-source tool that will provide cryptographic proof of model provenance.

By creating AI model ID cards with secure hardware and binding models to specific datasets and code, AICert aims to establish a traceable and secure LLM supply chain.

The proliferation of LLMs demands a robust framework for model provenance to mitigate the risks associated with malicious models and the spread of misinformation. The development of AICert by Mithril Security is a step forward in addressing this pressing issue, providing cryptographic proof and ensuring a secure LLM supply chain for the AI community.

(Photo by Dim Hou on Unsplash)

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The event is co-located with Cyber Security & Cloud Expo.

Explore other upcoming enterprise technology events and webinars powered by TechForge here.

The post Mithril Security demos LLM supply chain ‘poisoning’ appeared first on AI News.

Moreover, despite the developments that have been contrived, some retailers have not yet acknowledged the true possibilities of AI and what this entails. It is these retailers that need to question themselves: what do we want to accomplish with AI? What can AI really deliver – and what will this mean for our customers?

The Power of AI 

The opportunities to leverage AI and ML to improve retail operations are exponential for either online, in store or in the warehouse. However, before AI can be truly powerful in the sector, the hurdles of data quality and quantity must be addressed.

Even with the magnitude of data currently captured by many retailers, they repeatedly struggle with capturing and retaining enough of the ‘better stuff’ – historical data that is factual, complete and textual – to take full advantage of the benefits AI can put forward. This data is fundamental to make AI and ML models the best they can be. In the rush to grasp innovation, it is easy to overlook the challenges presented and move focus away from the essential objectives, which should be: where is the ROI, and what is the implication for the customer experience?

Take, for example, a great AI solution in store that gathers information about a customer – such as hair colour, size, and style – to recommend new products, from hair-styling to make-up, fashion to accessories. This could be extremely enticing for customers, especially due to the pandemic, as it allows for minimal contact and social distancing.

Future Goals

With the future of various high street stores remaining questionable, getting the customer experience right at every opportunity – which means encouraging repeated visits – is critical. In today’s highly competitive retail landscape, ridden by challenges from Covid-19, consumers continue to face seismic hurdles with both offline and digital worlds. Therefore, it’s crucial to understand the key benefits that new technology will bring, including the customer experience implications. 

Today, customers will seek different retailers within the overloaded omni-channel marketplace if products can be delivered efficiently as well as being at a lower price elsewhere. Strong customer experience and satisfaction are not achieved through technology gimmicks, but through ensuring technology can help customers quickly get their hands on the products they are searching for before it’s too late. 

In practice, this means ensuring maximum stock availability where it is needed, at the correct time, and at the right location. It means achieving a slick warehouse and distribution centre operation that can deliver both to the store and to the consumer’s front door whilst adhering to social distancing requirements. AI solutions will add positive value when this structure is in place.

Change is good

Retailers that have utilised AI in the warehouse are already driving tangible developments in efficiency and accuracy. By combining high-quality order history data with AI and ML to greater grasp the aspects of purchasing trends, including direct to consumer ecommerce orders, retailers can reconsider operations. Schedules can be re-organised and resources rearranged, while orders can be seamlessly prioritised and new delivery choices enabled. 

In addition, AI enables retailers to better manage changing sales peaks such as the back to school rush or a sudden April heatwave. The huge rise in ecommerce orders experienced when stores closed their doors last year, indicates how retailers need to be well prepared if a new peak suddenly emerges. 

With the help of AI, retailers can take advantage of accurate and granular predictions that can be utilised to smooth out the entire logistics process and efficiently respond to challenges in real-time. With innovation at the core of AI, retailers of all types and sizes can use the myriad of AI and ML opportunities to level the playing field with giant retailers.

AI is taking retail by storm, however, putting the gimmicks aside, the real and attainable value of AI currently is to be gained by applying proven algorithms to drive fundamental supply chain developments.

Find out more about Digital Transformation Week North America, taking place on November 9-10 2021, a virtual event and conference exploring advanced DTX strategies for a ‘digital everything’ world.

The post Utilising AI for retail in a post-pandemic world appeared first on AI News.