Almost every executive using artificial intelligence (AI) for professionals has faced at least one problematic incident, but only a handful of companies are using the new-age technology responsibly, Infosys said in a report on Thursday.
The ‘Responsible enterprise AI in the agentic era’ report by Infosys Knowledge Institute, the research arm of Infosys, found that 95% of executives who use enterprise AI experienced at least one incident. The report took inputs from 1,500 executives in the US, Canada, the UK, Germany, France, and Australia and New Zealand (ANZ).
Privacy violations, systemic failures, inaccurate or harmful predictions, and ethical violations were the most common incidents that executives sampled for the survey reported. Nearly three-quarters of the companies covered in the Infosys report considered the damage ‘substantial,’ while 39% said the impact was ‘severe’ or ‘extremely severe.’
Financial losses most common in AI incidents
More than three-fourths (77%) of the time, these incidents resulted in a direct financial loss, while the remaining instances involved reputational or legal damages, according to the Infosys report. However, executives considered loss of face much more threatening to their business than financial losses, it added.
Despite being the most common, the size of financial losses due to AI errors is relatively small, the report said.
“The average company in our sample reported financial losses from enterprise AI incidents of about $800,000 over two years. In total, this equates to between $750 million and $1.5 billion across the sample, and when extrapolated, represents an annual cost of between $1.4 billion and $2.9 billion globally across all businesses ($2.1 billion on average),” the report said.
Responsible AI on back burner
Despite the widespread troubles, the Infosys survey found only 2% of the surveyed companies meeting the IT services company’s standards of responsible AI use. Nearly 15% of the sample size meet three-fourths of the standards, while 83% implement them in a piecemeal manner, the report said.
Most executives cited the lack of resources and the ever-changing regulations for their weak RAI processes. On average, leaders sought an additional 30% of responsible AI (RAI) spending, which already accounts for 25% of overall AI costs. Meanwhile, financial losses from enterprise AI incidents amount to only 8%, making the demand for higher spending a risky premium.
On the human resources front, a larger RAI team enables the deployment of more enterprise AI initiatives. However, the success rate of deployments as a proportion of total initiatives falls from 24% to 21% as team size grows.
Independence Day 2025
Before Trump, British used tariffs to kill Indian textile
Bank of Azad Hind: When Netaji gave India its own currency
Swadeshi 2.0: India is no longer just a market, it’s a maker
Privacy violations, systemic failures, inaccurate or harmful predictions, and ethical violations were the most common incidents that executives sampled for the survey reported. Nearly three-quarters of the companies covered in the Infosys report considered the damage ‘substantial,’ while 39% said the impact was ‘severe’ or ‘extremely severe.’
Financial losses most common in AI incidents
More than three-fourths (77%) of the time, these incidents resulted in a direct financial loss, while the remaining instances involved reputational or legal damages, according to the Infosys report. However, executives considered loss of face much more threatening to their business than financial losses, it added.
Despite being the most common, the size of financial losses due to AI errors is relatively small, the report said.
“The average company in our sample reported financial losses from enterprise AI incidents of about $800,000 over two years. In total, this equates to between $750 million and $1.5 billion across the sample, and when extrapolated, represents an annual cost of between $1.4 billion and $2.9 billion globally across all businesses ($2.1 billion on average),” the report said.
Responsible AI on back burner
Despite the widespread troubles, the Infosys survey found only 2% of the surveyed companies meeting the IT services company’s standards of responsible AI use. Nearly 15% of the sample size meet three-fourths of the standards, while 83% implement them in a piecemeal manner, the report said.
Most executives cited the lack of resources and the ever-changing regulations for their weak RAI processes. On average, leaders sought an additional 30% of responsible AI (RAI) spending, which already accounts for 25% of overall AI costs. Meanwhile, financial losses from enterprise AI incidents amount to only 8%, making the demand for higher spending a risky premium.
On the human resources front, a larger RAI team enables the deployment of more enterprise AI initiatives. However, the success rate of deployments as a proportion of total initiatives falls from 24% to 21% as team size grows.
