CAPTCHA

Completely Automated Public Turing test to tell Computers and Humans Apart or CAPTCHA was invented in 2000 as a means to deter bots and spam on publicly available websites.^[1] CAPTCHA tests aim to confirm that the visitor of a website or service is human, usually by presenting a challenge which humans can solve easily, but computer programs cannot. Primary CAPTCHAs used today are Google's reCAPTCHA and hCaptcha.

"It's an arms race between site owners and spammers; users lose." - Jeremy Elson^[1]

According to a study by Searles et al., "...it can be concluded that reCAPTCHAv2 presents no real security.".^[2] Datadome found that half of passed CAPTCHAs were completed by bots.^[3] In response, the latest CAPTCHAs are using aggregated data of browsing history to rate a user's "humanness", presenting concerns around tracking and privacy.^[4]

The World Wide Web Consortium (W3C) releases a periodic report on the inaccessibility of CAPTCHA technology. Their 2021 report concluded that "traditional CAPTCHA continues to be challenging for people with disabilities, but also that it is increasingly insecure and arguably now ill-suited to the purpose of distinguishing human individuals from their robotic impersonators."^[5] It is important for websites to be able to keep unwanted bots from accessing their sites, however CAPTCHA may not be the best way to do so.

Newer forms of CAPTCHA work by scraping a user's device and behavior for uniquely identifiable information which would indicate a unique human using the service, as opposed to a bot which would have known and repetitive information. Information collected can include screen size, IP address, mouse and touch activity, previous websites visited, etc.^[6]

Services such as Google's reCAPTCHA have been found to be using human input to perform transcription work or train machine-learning models without user consent. On 22 January 2015, a Massachusetts class-action lawsuit attempted to argue Google should pay its users for their labor.^[7] Google's motion to transfer the case to the Northern District of California was granted 12 August 2015,^[8] where it was dismissed on 3 February 2016.^[9]

The W3C also outlined potential consumer-positive alternatives to CAPTCHAs:^[10]

1. Honeypot - "Another method to detect automated submissions. The idea behind the honeypot method is as follows: website forms would include a hidden field (by positioning the field off screen). Since spam robots cannot detect a hidden field in the HTML, when data is inserted into this 'honeypot' field, the website administrator would know that the data was not entered by a 'real' user."
2. Temporary tokens - after a user passes a CAPTCHA, a token is accepted onto the user's device allowing them to use the associated webservice for a fixed amount of time.
3. Multi-factor authentication - using a pre-arranged secondary device to independently authenticate identity.
4. Biometric authentication - facial recognition, fingerprint, retinal scan. This would only be acceptable in an institution with very high security requirements.

"Users should not be forced beyond what is strictly necessary to keep a site secure, e.g., if a honeypot suffices, use a honeypot until evidence of robotic attacks dictates something else." - W3C^[5]

• DataDome
• hCAPTCHA
• Privacy Pass
• reCAPTCHA