Since AI has been advancing at such a rapid pace, it stands to reason that negative AI incidents will also be on the rise with all things having been considered and taken into account. It turns out that 2023 was a record breaking year, with 121 incidents recorded according to a recent report released by Surfshark. This represents a 30% increase from 2022, and it also comprises a solid 20% of all AI incidents recorded since 2010.
With all of that having been said and now out of the way, it is important to note that OpenAI was involved in over 25% of the incidents that were factored in. Microsoft came in second with a total of 17 incidents, followed by Google with 10 and Meta with 5.
Quite a few of these incidents had to do with deepfakes and various forms of impersonation, with figures like Pope Francis, Tom Hanks and others becoming the subject of AI generated images. Politicians were also popular targets, with everyone from Donald Trump to Barack Obama being included in this list. 2024 is an election year, which is pertinent because of the fact that this is the sort of thing that could potentially end up making the number of incidents become even higher than might have been the case otherwise.
It bears mentioning that these incidents actually became somewhat less prevalent in the latter half of the year. The first quarter of 2023 saw 54 incidents, followed by 33 in Q2, but in the third and fourth quarters this plummeted to 14 and 22 respectively.
It will be interesting to see where things go from here on out. The downward trend seems to suggest that the perpetrators of AI incidents are losing interest, but in spite of the fact that this is the case, the election year might lead to a spike that will break even more records. Whatever the case may be, AI will continue to become ever more advanced. That will only make these incidents harder to detect or prevent in the first place, and they’ll also be far more accurate than they are right now.
Read next: IEA Projects Data Center Electricity Needs to Exceed 1,000 Twh by 2026, Raising Environmental Concerns
With all of that having been said and now out of the way, it is important to note that OpenAI was involved in over 25% of the incidents that were factored in. Microsoft came in second with a total of 17 incidents, followed by Google with 10 and Meta with 5.
Quite a few of these incidents had to do with deepfakes and various forms of impersonation, with figures like Pope Francis, Tom Hanks and others becoming the subject of AI generated images. Politicians were also popular targets, with everyone from Donald Trump to Barack Obama being included in this list. 2024 is an election year, which is pertinent because of the fact that this is the sort of thing that could potentially end up making the number of incidents become even higher than might have been the case otherwise.
It bears mentioning that these incidents actually became somewhat less prevalent in the latter half of the year. The first quarter of 2023 saw 54 incidents, followed by 33 in Q2, but in the third and fourth quarters this plummeted to 14 and 22 respectively.
It will be interesting to see where things go from here on out. The downward trend seems to suggest that the perpetrators of AI incidents are losing interest, but in spite of the fact that this is the case, the election year might lead to a spike that will break even more records. Whatever the case may be, AI will continue to become ever more advanced. That will only make these incidents harder to detect or prevent in the first place, and they’ll also be far more accurate than they are right now.
Read next: IEA Projects Data Center Electricity Needs to Exceed 1,000 Twh by 2026, Raising Environmental Concerns
