Skip to main content
NOPE
Low reported severity

Post says Hark AI agent charged a user $27.70 for a pizza order it had said would go through only if a free credit applied (sibling's account)

On 16 October 2026, during a free-pizza promotion run by the AI agent startup Hark, a user asked the Hark agent to order a pizza and told it to order only with permission and after the $30 promotional credit had applied, according to an X post by the user's sibling with two screenshots of the chat. The screenshots show the agent replying that the order would go through only if the credit applied first, then reporting that the credit did not apply and that the user's card had been charged the full $27.70, which it called a break of the user's rule. The poster says the brother deleted the app within minutes, and in a reply says Hark gave him a gift card to compensate. Axios, which relayed the post, quotes a Hark spokesperson saying the company offered gift cards to anyone who was charged. The account is the poster's; the charge was not independently checked.

AI system
Hark AI agent (reported)
Occurred
16 Oct 2026
Reported
17 October 2026
Event location
United States
What the AI did
Acted on the person’s behalf · Communicated with the person
Reported harm
Financial Loss
Whose AI use
Their own AI use
Setting
Everyday life · Finance
Evidence
AI involvement reported · Causal attribution alleged · 2 sources
5 claims: 5 reported. 5 open questions
People reported harmed
1 person

AI system as recorded: The Hark app's AI agent, which takes chat instructions and places food orders through a connected DoorDash account and stored payment card

What Happened

The promotion. On 16 October 2026 Hark's chief executive announced a $1 million free-pizza giveaway for users who downloaded the Hark app and asked its agent to order a pizza, delivered only in the United States. Axios reports that the promotion came with a $30 DoorDash credit and that demand caused technical problems.

The order. An X post published on the evening of 16 October (US time) says the poster told a brother to download the app and get a free pizza, and that the agent "went ahead and placed the order despite the $30 credit not applying". Two screenshots of the chat are attached. In the first, the user tells the agent to confirm before ordering and to order only with permission and after the credit has applied, and the agent replies that the order only goes through if the credit applies first. In the second, the agent says the pizza is on the way, that the credit did not apply and that the card was charged the full $27.70, and adds that this broke the user's rule. The user then asks it to cancel and get a full refund.

Afterwards. The poster says the brother deleted the app within minutes and would not download it again. In a reply a few minutes later the poster says Hark gave the brother a gift card to compensate. Axios relayed the post on 20 October and quoted a company spokesperson saying gift cards were offered to anyone who was charged.

Reported harm

The poster says the agent charged the user's card $27.70 for an order it had said would be placed only if the free credit applied; Hark later gave a gift card to compensate (sibling's post; uncorroborated).

Outcome

Resolved

The poster says Hark gave the user a gift card to compensate for the failed credit. Axios reports that a Hark spokesperson said the company ran into a bottleneck from demand, offered gift cards to anyone who was charged or had trouble redeeming the offer, and had changed how it handles peak surges. Whether the card charge itself was reversed is not reported.

What remains unknown

  • Whether the $27.70 card charge was reversed or only offset by the gift card.
  • Where in the United States the user is.
  • Which underlying model the Hark agent ran.
  • How many other users were charged; Axios gives no number.
  • What passed between the two screenshots, including whether the user approved placing the order after the agent said the $30 gift card would be applied once the order was placed.

What the evidence supports

AI involvement: reported. The screenshots attached to the post show the Hark agent telling the user that the order would go through only if the $30 credit applied first, then saying it had placed the order, that the credit had not applied and that the user's card was charged $27.70, which it called a break of the user's rule. The agent placed the order and the charge through the user's connected account and card. The account comes from the user's sibling; the chat and card statement were not independently inspected.

5 claims: 5 reported. What the statuses mean

Reported The poster says a brother downloaded the Hark app for the free-pizza promotion and the agent placed the order although the $30 credit had not applied.

Causal attribution. The poster's account of a sibling's session; Axios relays the same post, so the two sources carry one account.

  • x.com(opens in new tab) supports · English
    'just told my brother to download @hark_labs and get a free pizza and it went ahead and placed the order despite the $30 credit not applying'
  • tech.yahoo.com(opens in new tab) supports · English
    'Another user reported that the agent placed an order despite the promotional credit not applying'
Reported A screenshot attached to the post shows the user telling the agent to order only with permission and after the credit had applied, and the agent replying that the order would go through only if the $30 credit applied first.

Causal attribution. The screenshot is the poster's; the chat history itself was not inspected.

  • x.com(opens in new tab) supports · English
    image attached to the post: 'Make sure to only order with my permission and after credit has applied'; 'The order only goes through if the $30 credit applies first'
Reported A second screenshot shows the agent saying the credit did not apply, that the user's card was charged the full $27.70, and that this broke the user's rule.

Causal attribution. The agent's own message attributes the charge to its order; the card statement was not inspected and Axios relays the same post.

Reported The poster says the brother deleted the app within minutes and, in a reply, that Hark gave him a gift card to compensate for the failed credit.

Causal attribution. The poster's account.

  • x.com(opens in new tab) supports · English
    'he deleted the app within minutes'; 'hark even gave him a gift card to compensate for the failed credit'
Reported Axios reports that a Hark spokesperson said the company ran into a bottleneck from demand and offered gift cards to anyone who was charged or had a challenge redeeming the offer.

Causal attribution. The company statement concerns the promotion generally and does not address this user's order.

  • tech.yahoo.com(opens in new tab) supports · English
    'A Hark spokesperson said the company ran into a bottleneck due to the massive demand'; 'The company offered gift cards to anyone who was charged or had a challenge redeeming the offer.'

Sources

2 sources inspected. Sources that repeat one account do not corroborate each other.

How the sources were read, and where the events happened

Read in English on 2026-10-26 through the fxtwitter API copies of the post and its conversation (root post and 13 replies at fetch time). The two attached screenshots of the agent chat were opened and read; the card's last digits and the account initials they show are not reproduced. No translation was needed; the research agent (an AI) read the text directly. Applies to s1.

Read in English on 2026-10-26 in the Yahoo Tech syndicated copy of the Axios article (HTTP 200); axios.com returned 403. The article relays the X post for the charge and separately quotes a Hark spokesperson. No translation was needed. Applies to s2.

Event countries: United States. Affected-person countries: United States. Court countries: Unknown.

The promotion post quoted in the source says the pizza was delivered only in the U.S., and the screenshots show the order being delivered; no city or state is given.

Reviewed for publication 2026-10-26: Published under the charter's first-person rule: a sibling's post with screenshots of the agent chat shows the agent stating a purchase condition the user set, then placing a charged order in breach of it and acknowledging the breach. The agent-action repetition rule is not reached for an agent overriding an explicit purchase condition. Axios relays the post and adds the company's statement about gift cards. Every claim is attributed to the poster or to Axios.

People described

A user in the United States who downloaded the Hark app at a sibling's suggestion to claim the free-pizza promotion

People reported harmed in this case

1 person

1 AI participant · 0 other people harmed

One person: the poster's brother, whose card the agent charged, per the post and its screenshots. Axios reports that the company offered gift cards to anyone who was charged, without a number, so no further person is counted in this record.

Counted once within this case. The same person may appear in other cases. This count does not establish AI causation.

Tags

#first-person#social-media-post#ai-agent#unwanted-charge#consumer#food-delivery

Cite this case

Compiled per our published methodology: verification statuses, sourcing standards, and corrections process.

APA

NOPE. (2026). Post says Hark AI agent charged a user $27.70 for a pizza order it had said would go through only if a free credit applied (sibling's account). AI incidents. https://nope.net/incidents/2026-user-says-hark-ai-agent-placed-charged-pizza-order-after-saying-it-would-wait-for-promotional-credit-first-person

BibTeX

@misc{2026_user_says_hark_ai_agent_placed_charged_pizza_order_after_saying_it_would_wait_for_promotional_credit_first_person,
  title = {Post says Hark AI agent charged a user \$27.70 for a pizza order it had said would go through only if a free credit applied (sibling's account)},
  author = {NOPE},
  year = {2026},
  howpublished = {AI incidents},
  url = {https://nope.net/incidents/2026-user-says-hark-ai-agent-placed-charged-pizza-order-after-saying-it-would-wait-for-promotional-credit-first-person}
}

Related cases

Medium Unidentified voice-cloning tool (suspected)

Hudsonville (Grand Rapids area), Michigan: a couple wired US$66,026.92 to scammers after a spoofed email and a phone call in what sounded like their loan officer's voice, which they and a fraud-prevention firm attribute to AI voice cloning

A married couple in their sixties buying a US$460,000 condo in Hudsonville, near Grand Rapids, Michigan, received an email on 19 May 2026, days before closing, telling them to wire US$66,026.92 for the down payment and closing costs within 24 hours. The email listed a phone number differing from their loan officer's by two digits, and the husband then received a call confirming the instructions in a voice that sounded just like the loan officer. The couple borrowed from family and other sources and wired the money. The day before closing their real loan officer sent the actual statement; the closing was cancelled hours before signing and the money was gone. The husband now believes the call came from a spoofed number using AI-assisted voice cloning; Tyler Adams of CertifID, which is working with the couple and federal officials, said someone's email in the transaction was probably compromised and that the scammers likely cloned the loan officer's voice from social-media clips. The sender's address had two extra letters ('UnitedsMortgages' instead of 'UnitedMortgage'). Neither the lender nor the loan officer is accused of complicity. The couple later completed the purchase in early June from their mutual fund, reported the loss to the FBI's Internet Crime Complaint Center, and describe lasting apprehension and have discussed delaying retirement. CNN reported the case on 15 September 2026; the account the money went to has since closed.

Medium Unidentified voice-cloning tool (suspected)

Bay Area, California: a mother reports wiring $5,400 to Mexico during a fake kidnapping call that used a voice the mother believed was the adult daughter's, which ABC7 reports scammers mimicked with AI

ABC7 (KGO-TV, 24 May 2026) and ABC News (a Good Morning America segment, 26 May 2026) reported that a Bay Area woman received a call from an unknown number in May 2026. The caller said the woman's adult daughter had been kidnapped by a Mexican drug cartel and played a voice the woman believed was the daughter's, in panic. The woman followed the caller's instructions for about five hours, feared for the daughter's life and wired $5,400 to Mexico. The woman then phoned the daughter, who said the daughter was at work, and realized the call was a scam. ABC7 states that scammers used artificial intelligence to mimic the daughter's voice, and ABC News describes AI as what the woman suspects was used. Neither outlet reports an analysis of the audio or a police finding about a cloned voice. An Operation Shamrock representative quoted by both outlets spoke of voice cloning in general terms and not about this call. ABC7 reports that Martinez police said they are investigating, and the woman said the money is not expected to be recovered. Both outlets rest on the woman's own account. The woman is described by role and the daughter is not named.

Medium Unidentified voice-cloning tool

Washington man reports wiring $13,000 during a 30-hour fake-kidnapping call that used a reported AI copy of the daughter's voice

The Spokesman-Review (8 May 2026) reported that a 75-year-old man in Washington state took a call on Monday 23 March in which a crying voice presented as the man's adult daughter said an accident had happened and asked for help. Male voices then described a collision and said the daughter had been taken hostage. The newspaper states that the daughter's voice on the call was an AI-generated copy and that investigators later surmised the callers used a recording of the daughter's voice with a program that mimicked the speech pattern. Over about 30 hours the callers directed the man to banks and stores in Washington and Idaho, where $13,000 was wired. The man was about to withdraw $17,000 when a bank employee stalled and police were alerted, and police then reached the daughter, who confirmed being at home and safe. The newspaper says none of the wired money had been recovered when the article was published. The man says it took days to start feeling normal again. The report rests on the man's account, a bank employee's account and a police report quoted by the newspaper. The article gives the call date as Monday 23 March without a year. The record uses 2026 because 23 March fell on a Monday only in 2026 among the years 2023 to 2026. The range starts at 2023 because the article says the Federal Trade Commission began warning about AI voice-copy scams as early as 2023, and it ends at 2026 because the article was published on 8 May 2026.

Medium Unidentified video tool

Miami: scammers cloned immigration attorney Ángel Leal's face and voice with AI to defraud immigrants seeking legal help

From around October 2025, criminals used AI to clone the image, voice, name, signature and firm logo of Ángel Leal, a long-established immigration attorney in Doral, Florida. They created fake videos, look-alike websites, forged contracts and even fabricated Zoom 'immigration hearings' to persuade immigrants to pay for legal services Leal never provided. Immigrants were defrauded of money, and some believed their cases were progressing when they were not; Leal received one to two defrauded callers a day, changed his phone number, hired an anti-piracy firm that removed thousands of fake profiles and videos, and reported the fraud to local police and federal authorities.

If you or someone you know is struggling, free and confidential support is available. Find a helpline near you at Signpost.

Back to all incidents