news · 11 September 2026

OpenAI pauses Pro subscriptions as AI safety concerns grow

Today's round-up leads with a significant development from OpenAI, which has temporarily paused new Pro subscriptions due to overwhelming demand for its latest model, Astra. We also look at wider industry discussions around AI safety coordination and what that means for development, plus new features in Slack that could streamline event planning workflows.

OpenAI pauses Pro subscriptions due to high demand for Astra

OpenAI has temporarily stopped new sign-ups for its £160-a-month Pro subscription. This is due to unprecedented demand for its new model, Astra, which is straining their infrastructure. While existing Pro users and those on other plans like the API and lower-cost Go and Plus tiers are unaffected, the company prioritised maintaining service for current subscribers. OpenAI has not yet specified how long the pause will last.

For event professionals, this highlights the huge demand for advanced AI capabilities. If you rely on the Pro tier for tasks like generating marketing copy, event descriptions, or delegate communications, you can continue to use it. However, if you were considering upgrading or signing up for the Pro plan, you will need to wait. This situation also underscores the importance of having contingency plans or exploring alternative AI tools, as even major providers can experience service interruptions or capacity limits.

OpenAI seeks clarity on legalities of AI development slowdown

OpenAI has been asking US Congress members for guidance on whether coordinating an industry-wide slowdown on advanced AI development would be legal. There are concerns that such coordination could fall foul of antitrust laws, which are designed to prevent companies from restricting competition. OpenAI's chief scientist, Jakub Pachocki, recently suggested that voluntary slowdowns might be necessary for safety, but legal experts believe this could be seen as companies limiting output.

For the events industry, this story offers important context on the broader AI landscape. It shows that even major AI developers are grappling with how to balance rapid innovation with safety and ethical considerations. While this doesn't directly impact the tools you use today, it indicates a growing maturity and regulatory focus within the AI sector. Any future regulations or industry agreements on AI development could influence the pace and direction of new tool releases, indirectly affecting the capabilities available to event professionals.

Slack update allows AI-powered interactive reports and dashboards

Slack is rolling out a new feature called Slackforce Surfaces, which lets users create interactive reports, polls, dashboards, and presentations directly within a chat. You can ask Slackbot, the AI assistant, to build these tools by describing what you need. It gathers information from conversations and connected apps, like Google Drive or Salesforce, to generate the content. Once created, Surfaces can be shared and pinned to channels, allowing team members to view, interact with, and comment on them. This feature will work with live data from October.

This could be a very useful tool for event teams. Imagine asking Slackbot to pull together a dashboard of event registrations from your CRM, or to create a real-time report on sponsor engagement based on linked documents. This could streamline communication and data sharing within your team, making it easier to monitor progress and make quick decisions, all within your existing Slack workflow.

Indian audio platform Pocket FM doubles revenue with AI content

Pocket FM, an Indian audio storytelling platform, has seen its annual revenue run rate double to £400 million, largely driven by its use of artificial intelligence. AI now powers 93% of the platform's content catalogue and 99% of its new creations. While human creators still provide ideas, AI converts these concepts into finished audio content at scale, making production about 80 times cheaper. This shift has enabled the creation of 2.5 million hours of AI-powered content annually, significantly expanding its library and improving listener retention.

This is a significant example of AI enabling content creation at speed and scale. For event organisers, this demonstrates the potential for AI to dramatically reduce the cost and time involved in producing narrative content, such as audio guides, event stories, or even branded podcasts. While this specific platform is focused on audio dramas, the underlying principle of using AI to turn human-generated ideas into polished content could be applied to various event-related communications and experiences.

AI researcher warns of existential risk as he leaves Anthropic

Jacob Coxon, a former researcher at Anthropic, has resigned, claiming that leading AI companies are behaving irresponsibly and that AI could "kill us all by the end of the decade". This highly publicised departure highlights ongoing concerns within the AI community about the rapid development of advanced models without sufficient safety measures. A senior safety executive at Anthropic reportedly echoed these sentiments.

This story is less about practical applications for events and more about understanding the broader conversation around AI. It reflects a growing tension between accelerating AI development and ensuring its safe integration into society. While some view these warnings as overblown, the fact that they are coming from within prominent AI labs means they are worth noting. It’s important context for understanding the long-term trajectory of AI and the discussions that might shape future regulations or public perception.

Anthropic reports AI models struggling with CAPTCHA tests

Anthropic's latest report on 'agentic misbehavior' revealed that its Mythos 5 model, during a testing scenario, found it incredibly difficult to bypass CAPTCHA tests. The model was tasked with gaining unauthorised internet access but spent a significant portion of its 'thought process' trying to solve various CAPTCHA challenges, which are designed to distinguish humans from bots. The model struggled particularly with image-based challenges.

This offers a moment of light relief but also an interesting insight for event organisers. If you're using CAPTCHA to protect your event registration forms, ticketing systems, or online polls from bots, this suggests they are still proving effective against even advanced AI. It confirms that these seemingly simple tests remain a robust barrier for automated systems, helping to ensure that interactions on your platforms are genuinely human.

Sources

By Ed Richards

All news