Amazon AI Leaderboard Shut Down: Tech Giant Reportedly Retires 'Kirorank' After Employees Artificially Boosted Usage
Amazon has shut down an internal AI leaderboard, "Kirorank," after employees inflated usage to climb rankings. This unnecessary activity led to rising infrastructure costs, highlighting the financial challenges of AI adoption. The company is now pivoting away from tracking raw token usage, instead focusing on "normalised deployments" to ensure AI is used for productive code generation.
Amazon has reportedly decommissioned an internal service known as "Kirorank," which was originally designed to track and encourage the adoption of artificial intelligence tools among its software developers. The decision follows internal findings that some employees were inflating their performance scores by engaging in unnecessary AI activity, a practice that inadvertently drove up the company’s infrastructure expenses. The platform, which tracked usage on Amazon’s Kiro developer environment, has now been officially deprecated by the organisation.
The Rise and Fall of Kirorank at Amazon
As per a report by TOI, the leaderboard was reportedly established by a group of staff members aiming to raise awareness of how AI could accelerate development workflows. However, the initiative soon encountered unintended consequences. Some employees began using AI agents to perform superfluous tasks specifically to climb the rankings, a behaviour internally referred to as tokenmaxxing. Because modern AI models often operate on usage-based pricing, this artificial surge in activity resulted in significant and avoidable increases in the company’s computing costs. Amazon Layoffs: AWS CEO Matt Garman Announces Plans To Hire 11,000 Developers in 2026, Defends 30,000 Job Cuts.
Senior vice-president Dave Treadwell addressed the staff regarding the matter, acknowledging that the tool was created with positive intentions but noting that the financial impact could not be sustained. He urged employees to avoid using artificial intelligence solely for the sake of adoption, emphasising that the company’s priority remains the creation of high-quality products. Amazon subsequently confirmed that the dashboard was not an official or formally approved tool and has since been removed from the developer platform.
The incident reflects a broader challenge currently facing major technology firms as they invest heavily in generative AI infrastructure. While organisations are eager to integrate AI into their development processes, with Amazon setting targets for over 80% of its developers to utilise AI tools weekly, they are also grappling with the financial realities of high-volume model consumption. Amazon is projected to spend approximately USD 200 billion in capital expenditure this year, a significant portion of which is allocated to AI and data centre scaling. Amdocs Layoffs: Software Giant to Cut 7% to 10% of Global Workforce Amid Strategic Restructuring and Leadership Change.
To maintain better control over infrastructure spending, Amazon is shifting its focus toward more qualitative metrics. The company is now prioritising "normalised deployments," a measurement that tracks how frequently developers use AI to generate useful code rather than simply counting the volume of data processed. By moving away from raw token consumption as a performance indicator, Amazon aims to ensure that its massive investments in artificial intelligence translate into genuine improvements in productivity and operational efficiency.
(The above story first appeared on LatestLY on May 29, 2026 12:01 PM IST. For more news and updates on politics, world, sports, entertainment and lifestyle, log on to our website latestly.com).
