The Weekly #3

UK Signs International AI Treaty

The UK has signed a groundbreaking international treaty on artificial intelligence aimed at preventing its misuse and safeguarding human rights. The Council of Europe treaty mandates safeguards against AI threats and requires transparency from both public and private sectors in their AI use. It marks a significant step towards establishing global standards for responsible AI development and deployment. The treaty was also signed by the US, the EU and Israel.

Plain AI is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

This treaty is in addition to the EU AI act it signifies a crucial step towards establishing a global framework for the responsible and ethical use of AI. As AI technologies continue to advance rapidly, this agreement underscores the importance of proactively addressing potential risks while still taking advantage of the benefits that AI brings.

Anthropic Goes Enterprise

On Sep-04, Anthropic announced a new Claude for Enterprise Plan. Generative AI use cases have been slow to take off at a corporate level with some of the main reasons being data security and access controls.  With Claude for Enterprise, Anthropic are meeting some of those challenges by providing business with greater controls on how their staff access Generative AI tools.  The new plan provides a number of interesting areas:

• Single sign-on (SSO) and domain capture: Securely manage user access and centralize provisioning control.
• Role-based access with fine-grained permissioning: Designate a primary owner for your workspace to enhance security and information management.
• Audit logs: Trace system activities for security and compliance monitoring. Audit logs will be available in the coming weeks.
• System for Cross-domain Identity Management (SCIM): Automate user provisioning and access controls. SCIM will also be available in the coming weeks.

Of course Open AI has offered an enterprise plan for a while, so it’s a positive move to see other companies follow suit and push the whole sector forwards.

Hackathon

This week, my colleagues in the Customer Success formed into groups to participate in a hackathon.  On the whole, we are not a technical group of individuals but we certainly know our way around the platform and understand the concepts of AI.  As a consequence, we had some really interesting use cases presented, all the way from being able to send out personalised emails based on updates to a dataset, to predicting future contract amounts based on our historic upsell and renewal data, enriched with other customer metrics we track across the business.

Regardless of what the individual project was, the one thing we all had in common was that preparing datasets is still the biggest part of a project. I talk about this in my “Demystifying AI: Part 2” post, but research shows that it can be around 60% of a project, and based on our experience this week, I would certainly agree with that.

Plain AI is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.