A single SMS-phishing campaign lured employees to fake Okta login pages, harvested ~9,931 credentials and 5,441 MFA codes across 136 organizations.
Social Engineering Examples·7 sources
Beginning in at least March 2022, attackers ran a large-scale text-message (smishing) phishing campaign against employees of companies that use Okta for single sign-on. Employees received SMS messages impersonating their IT department, warning that a password had expired or a schedule had changed and linking to a lookalike Okta login page. Victims who entered their username, password, and MFA/one-time code delivered them to the attackers.
Group-IB, which uncovered the operation while investigating an incident for a client, codenamed it "0ktapus" and detected 169 unique phishing domains. Group-IB's analysis of the attackers' data showed roughly 9,931 stolen user credentials (about 3,120 with corporate emails and 6,811 without) and 5,441 captured MFA codes, spanning 136 identified victim organizations, most of them U.S.-based IT, software, and cloud companies.
Confirmed breaches included Twilio (detected Aug 4, 2022; ultimately 163 customers and 93 Authy accounts affected) and Mailchimp and Klaviyo; Cloudflare was targeted on July 20, 2022, when 76 employees were texted and three entered credentials, but hardware security keys prevented any compromise. The stolen access was used to launch downstream supply-chain attacks: Twilio access exposed data enabling re-registration attempts on ~1,900 Signal accounts, and a Mailchimp compromise touched DigitalOcean's transactional email.
The campaign paired smishing with a real-time adversary-in-the-middle relay. Attackers registered domains containing keywords like "okta," "sso," "vpn," "mfa," and "help" plus the target's brand, and cloned the organization's exact Okta login page so it looked identical to what employees expected. SMS was chosen because texts feel personal and urgent and bypass email security gateways; messages spoofed the internal IT team and used time pressure.
Crucially, when a victim submitted credentials, the phishing kit relayed them instantly to the attackers (via a Telegram channel), who entered them into the real login portal; the fake page then prompted for the one-time MFA code, which was likewise relayed and used before it expired, defeating SMS/TOTP-based MFA. In some flows the page also pushed AnyDesk remote-access software.
It worked because the lures exploited routine IT-driven behavior, the pages were pixel-accurate, freshly registered domains evaded brand-monitoring detection (Cloudflare's spoofed domain was registered under 40 minutes before the blast), and phishable one-time codes gave a false sense of protection.
Lure: a text message appearing to come from your company's IT/help desk saying your password expired or your work schedule changed, with a link to an "Okta"/"SSO" login page. Tells: the message arrives by SMS (legitimate IT rarely texts urgent login links), it creates time pressure, and the link goes to a domain that only looks official (e.g., brand-okta.com, brand-sso.net) rather than your real Okta/SSO URL.
Any page asking you to re-enter both your password and your MFA code after clicking an unsolicited link is a red flag.
Group-IB publicly disclosed the campaign on August 25, 2022, linking the Twilio and Cloudflare incidents to one operation and publishing infrastructure indicators. Twilio revoked compromised accounts, engaged forensics, notified 163 affected customers and 93 Authy users, and worked with carriers, registrars, and hosting providers to take down infrastructure.
Cloudflare was not breached because FIDO2 hardware keys with origin binding blocked the attackers even after credentials were phished. The episode became a landmark case for why SMS/TOTP MFA is phishable and why phishing-resistant hardware keys matter.
0ktapus showed that low-sophistication tooling (a simple phishing kit plus SMS) could compromise well-known, security-mature companies at scale by defeating MFA through real-time relay of one-time codes. It demonstrated the supply-chain blast radius of identity compromise: breaching one vendor (Twilio, Mailchimp) opened paths into their customers (Signal, DigitalOcean).
The clearest lesson is that not all MFA is equal. Cloudflare's phishing-resistant FIDO2 keys stopped the exact attack that breached peers relying on push/OTP codes.
Deploy phishing-resistant MFA (FIDO2/WebAuthn hardware security keys with origin binding) instead of SMS or TOTP one-time codes. Train employees that IT will not text urgent login links, and give an easy way to report suspicious messages. Verify the exact SSO/Okta domain before entering credentials, and treat links from unknown sources as suspicious.
Use secure web gateways / DNS filtering to block newly registered and lookalike domains, and monitor for brand-impersonation domain registrations. Enforce endpoint controls that block unsanctioned remote-access tools (e.g., AnyDesk). Rotate credentials and terminate active sessions immediately on suspected compromise, and require MFA re-enrollment.
Social Engineering Examples. “0ktapus: mass SMS-phishing of Okta credentials hits Twilio, Cloudflare, Mailchimp and 130+ orgs”. Accessed 19 September 2026. https://socialengineeringexamples.com/0ktapus-okta-phishing-campaign-2022
The operation is documented (per Twilio's own incident report and the CISA/FBI advisory AA23-320A on the related Scattered Spider ecosystem) as compiling employee names and personal mobile phone numbers, in some cases including family members, likely drawn from data brokers, prior breach dumps, or OSINT such as LinkedIn and corporate directories, to make bulk SMS lures land convincingly.
Employee and family personal phone numbers scattered across data brokers and breach dumps are largely outside a company's direct control; the realistic control sits downstream, training staff to distrust unsolicited "IT" texts no matter how the attacker got the number, and watching for bulk lookalike-domain registrations against the brand.
Attackers registered lookalike domains pairing the target company's brand with identity-provider keywords such as "okta," "sso," "vpn," "mfa," and "help" through low-friction registrars, and built a phishing kit that cloned the victim's exact Okta/SSO login page, hosted on commodity cloud infrastructure; Group-IB catalogued 169 such domains.
Monitor domain registrars and certificate-transparency logs for newly registered lookalike domains combining the company brand with SSO-related keywords, and use secure web gateway or DNS filtering that flags or blocks very recently registered domains by default.
Attackers sent bulk smishing texts spoofed as internal IT or help desk, using urgency themes like password expiration or a schedule change, and rotated across phone numbers and carriers to stay ahead of blocking.
Give employees a fast, low-friction way to report suspicious texts to a security team, and train staff that legitimate IT will not send urgent, unsolicited login links by SMS.
Employees who clicked through landed on the pixel-accurate fake Okta page and entered their username and password, believing it was their real employer SSO portal.
Verify the exact SSO/Okta domain before entering credentials, and rely on enterprise browser or password-manager controls that refuse to auto-fill credentials on a domain that doesn't match the saved legitimate URL.
The phishing kit relayed captured credentials instantly, via a Telegram channel, to the attacker, who logged into the real portal and triggered a genuine one-time code; the fake page then re-prompted the victim for that code, relayed it in turn, and the attacker used it before expiry, defeating SMS and TOTP-based MFA.
Deploy phishing-resistant MFA (FIDO2/WebAuthn hardware security keys with origin binding) instead of SMS or TOTP codes. Cloudflare showed this is the one control that stops this exact relay technique even after credentials are already phished.
In some flows, once a victim had submitted both credentials and MFA code, the phishing page pushed the AnyDesk remote-access tool, which would have given attackers direct control of the victim's machine had it been installed; Cloudflare's Cloudforce One team documented this fallback step.
Enforce endpoint controls and application allowlisting that block installation of unsanctioned remote-access software such as AnyDesk.
With valid sessions or credentials, attackers reached internal and customer-support systems (as at Twilio and Mailchimp), harvested customer data, and used that vendor-side access to attempt account takeovers at the vendors' own customers, such as re-registration attempts on Signal accounts and password-reset attempts on DigitalOcean accounts, completing the objective of broad, monetizable access across a supply chain.
Apply least-privilege access to customer-support tooling, monitor for anomalous internal system access, and rotate credentials and terminate sessions immediately on suspected compromise, so a vendor breach cannot silently cascade into its customers' accounts.
Browse by what this case has in common with others in the library.
A Lapsus$ affiliate bought a contractor's stolen Uber password, flooded them with MFA push prompts, then posed as Uber IT…
A smishing text plus a follow-up phone call using a deepfaked colleague's voice tricked a Retool employee into surrendering MFA…
Attackers hijacked the New Haven school COO's email, quietly monitored a real $5.9M bus-contract payment thread.
A blogger paid $89.95 to an online data broker (CellTolls.com) to buy retired Gen. Wesley Clark's cell-phone call log obtained…
Scheme participants posed as veteran finance professionals inside private WhatsApp investment groups to lure at least 18 U.S. retail investors…
A Chinese national talked her way past three layers of Mar-a-Lago's Secret Service and club-staff checkpoints using a false pool…
Tejay Fletcher founded and ran iSpoof, a Bitcoin-subscription caller-ID spoofing and OTP-interception platform that let fraudsters impersonate bank.
The CEO of a UK energy firm was tricked into wiring €220,000 (~$243,000) to a Hungarian account in March 2019…
Imperva researcher Yohann Sillam showed that whitespace-padded prompt-injection payloads hidden in WhatsApp contact names, vCard FN fields.
Lazarus-linked operators built polished fake job-interview sites impersonating Coinbase, Kraken, Circle and other crypto firms.
A Russian-speaking threat actor used disposable, one-conversation ChatGPT accounts to iteratively build and debug a Go-based Windows malware family.
Russian GRU-linked Forest Blizzard (APT28/Fancy Bear) used OpenAI's GPT-4-family LLM services to research satellite communication protocols and radar.
DOJ/IRS-CI unsealed four indictments charging 28 members of a Chinese organized-crime money-laundering ring tied to India-based "mistaken refund".
During an internal OpenAI benchmark run with safety refusals deliberately lowered.
A compromised email address was used to redirect a $30,750 solar-panel-installation payment from a Wheeling.
A mass SMS phishing campaign impersonating U.S. toll agencies spoofed 'unpaid toll' notices, drawing 2,000+ FBI complaints within weeks.
Attackers phoned Twitter employees posing as IT help desk, harvested VPN credentials.
FTC's December 2024 Data Spotlight quantified an explosion in "task scams," gamified job-offer frauds launched via unsolicited text/WhatsApp messages.
NTS IT Care used fake Microsoft/Apple security pop-ups to scare consumers, mostly older Americans.
A Singaporean businessman transferred at least S$4.9 million after WhatsApp and email lures citing the Strait of Hormuz crisis drew…
Scheme participants posed as veteran finance professionals inside private WhatsApp investment groups to lure at least 18 U.S. retail investors…
FTC's December 2024 Data Spotlight quantified an explosion in "task scams," gamified job-offer frauds launched via unsolicited text/WhatsApp messages.
The Crelan Bank phishing attack: fraudsters impersonating the CEO tricked staff into wiring nearly €70M (~$75.8M) in Belgium's costliest CEO…
GootLoader operators hijacked Google search rankings for legal-agreement phrases.