Digital briefing, October 10: Oracle's ChatGPT workflows, AI summaries in brand searches
OpenAI's customer success stories, the rapid spread of AI summaries in Google search, and Claude's new rules against fake source networks.
- Automation
Oracle says it cut business processes from days to minutes using ChatGPT and Codex #
OpenAI announced that Oracle is using ChatGPT Work and Codex across hiring, engineering and operations to turn expert knowledge into fast workflows. This was shared on OpenAI's own site; the full text of the page wasn't accessible, so only the summary is available.
What it means for your business: Large companies automating repetitive work with AI is a sign that similar tools may become accessible to smaller businesses over time.Source: OpenAI — How Oracle turns days of work into minutes with ChatGPT and Codex
- Search
Most searches for big brand names on Google now return an AI summary #
In Ahrefs' weekly tracking of 100 major brand name searches between July and September, only 5 results showed an AI Overview on July 1; by September 30 that number had risen to 63, with results published on October 2. Separate measurements by DemandSphere and Chris Long, using their own data samples, showed a similar increase, though the exact figures differed.
What it means for your business: Business owners can search their own brand name on Google to check the AI summary that appears and which sites are cited as sources, giving them a view into how their brand is represented in search results. - AI
Asana cut costs 76-fold while testing its browser assistant with a new OpenAI model #
OpenAI announced that Asana used its new model via Codex to test its browser-based AI assistant, cutting costs 76-fold and increasing speed 5-fold in these tests. This was shared on OpenAI's own page; the full text of the page wasn't accessible.
What it means for your business: The rapidly falling cost of AI-powered tools suggests these kinds of assistants may become affordable for smaller-budget businesses in the near future.Source: OpenAI — Asana cuts model costs 76x in browser tests with GPT-6.1 Sol
- Automation
Sophos cut cyber threat investigation time by 96% using OpenAI's Daybreak tool #
OpenAI announced that cybersecurity company Sophos used a tool called Daybreak to cut threat investigation time by 96% and automate 52% of MDR (managed detection and response) cases, while maintaining human oversight. The full text of the page wasn't accessible, so only the summary is available.
What it means for your business: It shows that even in a field as specialized as cybersecurity, AI can speed up routine analysis work, which could mean faster response times for businesses that use security services.Source: OpenAI — Sophos cuts threat investigation time by 96% with OpenAI Daybreak
- AI
Anthropic bans using Claude to build fake source networks, effective November 12 #
In its updated usage policy published on October 8, Anthropic explicitly banned using Claude to feed search engines or AI systems with fake or seemingly independent sites that conceal their origin; the rule takes effect on November 12. The company said that in September it identified and shut down an account linked to a France-based ad agency that had used Claude to produce at least 8,913 articles in 20 languages across roughly 70 fake local news sites.
What it means for your business: For businesses using AI in content production, it's important to know that building networks of fake or seemingly independent sites is now explicitly banned by Anthropic, and accounts doing so can be suspended.