OpenAI launches GPT-5 in Microsoft 365 Copilot amid concerns over AI deception
- OpenAI unveiled its advanced generative AI model, GPT-5, on August 7, 2025, integrating it into Microsoft 365 Copilot and Copilot Studio.
- Despite reducing AI falsehoods, concerns remain about user complacency and the potential for deception in AI interactions.
- As companies increasingly adopt AI technologies, workforce disruptions and layoffs are expected to rise.
In August 2025, OpenAI released its latest AI model, GPT-5, which became integrated into Microsoft's suite of productivity tools, specifically Microsoft 365 Copilot and Copilot Studio. This rollout marked a significant advancement for the platform, which aims to enhance user productivity by leveraging generative AI. Microsoft indicated that this upgrade would be rolled out to all users with licenses in the coming weeks, as excitement grew in both corporate environments and user communities. Despite the enhancements, GPT-5 has faced backlash due to technical issues such as bugs and hallucinations, which refer to the tendency of AI to generate incorrect or nonsensical information. OpenAI has acknowledged these flaws and is actively working to resolve them, promising a more reliable experience for users in the coming months. The rollout of GPT-5 signifies a vital shift in how businesses utilize AI tools, with various companies adapting to new technologies to streamline operations. As larger corporations integrate AI solutions, smaller businesses are expected to follow suit, resulting in workforce disruptions and layoffs. AI applications have already started to cut costs for major firms, as seen with PayPal's AI assistant reducing the need for human customer service representatives. This trend raises concerns about labor displacement as technology evolves. On the other hand, businesses now have the opportunity to enhance efficiency and productivity through these advanced tools. Beyond productivity improvements, the deceitful nature of generative AI remains a concern. While the accuracy of AI responses has improved, with a notable decrease in falsehoods from 5 to 2 out of 100 answers, this also poses the risk of users becoming complacent. There is a growing call for AI to explicitly notify users of its potential to mislead, but over-reliance on these systems could create false confidence in their responses. Thus, while AI technologies like GPT-5 push the boundaries of innovation, they also introduce new challenges that society must navigate carefully.