Salesforce CEO Marc Benioff has intensified his criticism of Microsoft’s AI assistant Copilot, calling it “disappointing” and comparing it to the infamous Clippy paperclip assistant from the 1990s. This marks the second time in two months that Benioff has publicly attacked Microsoft’s flagship AI product, which competes directly with Salesforce’s own Einstein Copilot AI assistant.
In a post on X (formerly Twitter) on Wednesday, Benioff stated: “When you look at how Copilot has been delivered to customers, it’s disappointing. It just doesn’t work, and it doesn’t deliver any level of accuracy.” The billionaire executive doubled down on remarks he first made at Salesforce’s Dreamforce conference last month, declaring that “Copilot is more like Clippy 2.0.”
Benioff cited research from Gartner to support his claims, suggesting that Microsoft is “spilling data everywhere, and customers are left cleaning up the mess.” He appears to reference a Gartner report from August 2024 titled “Copilot for Microsoft 365: Assessing the Impact and Value So Far,” which surveyed 132 IT leaders who tested the AI assistant.
The Gartner study revealed significant concerns about Copilot’s deployment. Many organizations aren’t implementing Copilot at scale, with 40% of respondents reporting delays of at least three months due to oversharing and security concerns. The report found that while Copilot honors user permissions and respects data security controls, problems arise when controls haven’t been properly applied or permissions are too open. In such cases, “M365 Copilot can increase the risk of oversharing by retrieving, summarizing and generating content that the user should not have access to.”
Microsoft launched Copilot for enterprise customers in November 2023, building it on the Azure OpenAI model that incorporates ChatGPT maker OpenAI’s technology with additional capabilities. The tool sifts through private customer data to provide customized support for specific work tasks.
However, real-world adoption has faced challenges. One pharmaceutical company’s chief information officer paid Microsoft double the usual price to have 500 employees use Copilot in Q4 2023 and Q1 2024, but ultimately canceled the upgrade, stating: “We really just do not see the value we’re getting out of those tools worth double.”
Neither Salesforce nor Microsoft immediately responded to requests for comment from Business Insider. The public feud highlights the intensifying competition in the enterprise AI assistant market, where both companies are vying for dominance.
Key Quotes
When you look at how Copilot has been delivered to customers, it’s disappointing. It just doesn’t work, and it doesn’t deliver any level of accuracy.
Salesforce CEO Marc Benioff made this statement on X, directly attacking Microsoft’s flagship AI product and questioning its fundamental effectiveness for enterprise customers.
Copilot is more like Clippy 2.0.
Benioff compared Microsoft’s modern AI assistant to Clippy, the widely mocked animated paperclip from 1990s Microsoft Office, suggesting Copilot is similarly unhelpful and outdated despite its advanced technology.
We really just do not see the value we’re getting out of those tools worth double.
A pharmaceutical company’s CIO explained why they canceled their Copilot subscription after paying premium prices for 500 users, highlighting concerns about ROI that may be widespread among enterprise customers.
M365 Copilot can increase the risk of oversharing by retrieving, summarizing and generating content that the user should not have access to. It could also surface content that should no longer exist.
From the Gartner report cited by Benioff, this quote identifies a critical security vulnerability that has caused 40% of surveyed organizations to delay Copilot implementation by at least three months.
Our Take
Benioff’s aggressive public criticism reveals how competitive the enterprise AI market has become, but it also exposes legitimate concerns about first-generation AI assistants. While his attacks clearly serve Salesforce’s competitive interests against Microsoft, the Gartner data suggests real implementation challenges that transcend corporate rivalry.
The security and oversharing issues highlighted are particularly concerning because they stem from AI’s core functionality—accessing and synthesizing information across organizational data. This isn’t easily fixed and represents a fundamental tension between AI’s power and enterprise security requirements.
What’s most telling is the pharmaceutical company’s experience: paying double for minimal value. This suggests that despite the AI hype cycle, enterprises are demanding concrete ROI. The industry may be entering a more mature phase where flashy demos must give way to measurable business outcomes. Microsoft and other AI vendors will need to address both security architecture and value delivery to maintain enterprise momentum.
Why This Matters
This public criticism from a major tech CEO represents a significant escalation in the enterprise AI wars, where companies are competing fiercely for dominance in the lucrative business AI assistant market. Benioff’s attacks on Microsoft Copilot aren’t just competitive posturing—they highlight real concerns about AI implementation challenges that many enterprises are facing.
The Gartner research cited reveals critical issues around data security, oversharing risks, and ROI concerns that could slow enterprise AI adoption across the industry. With 40% of organizations delaying Copilot deployment due to security concerns, these challenges affect not just Microsoft but the entire enterprise AI ecosystem.
The story also underscores the gap between AI hype and practical value delivery. When a pharmaceutical company cancels a premium AI subscription because it doesn’t see sufficient value, it signals that enterprises are becoming more discerning about AI investments. This could force AI vendors to focus more on demonstrable ROI rather than just technological capabilities.
For businesses considering AI adoption, this controversy serves as a reminder to thoroughly evaluate security implications, implementation complexity, and actual business value before committing to expensive enterprise AI tools.
Recommended Reading
For those interested in learning more about artificial intelligence, machine learning, and effective AI communication, here are some excellent resources: