Follow
Subscribe via Email!

Enter your email address to subscribe to this platform and receive notifications of new posts by email.

What is Phishing and How Do Attackers Steal Credentials?

Phishing is a deceptive social engineering attack where cybercriminals use fake emails, bogus websites, and phone calls to steal credentials, hijack sessions, and defraud individuals and organizations.
Cover page of Security Journal special issue on human factors with dark teal background

Understanding what is phishing begins with seeing how online trickery plays on human trust. Phishing is a social engineering scam where bad actors mimic real firms or known people to lure targets into giving up passwords, handing over cash, or loading bad code onto a computer. How do fake messages trick wary users into opening dangerous links? They build false urgency through spoofed warnings and bogus sign-in alerts [1]. Intruders bypass hard math and firewalls by nudging human choices, gaining entry to email and file tools like Google Workspace and Microsoft 365 [2].

What Is Phishing?

At its root, phishing covers fake messages built to trick people into sharing secret data or running rogue code. Criminals send these scams through email, fake websites, text chats, and voice calls. Their core goal is simple: slip past security barriers and steal private data [1].

Thieves chase clear financial goals when sending online lures [1]. Many plots seek user names, passwords, and credit card numbers for fast cash theft [2]. Crooks also trade stolen logins on darknet forums, selling bundles of passwords and cheap access to streaming accounts. Once inside an account, an attacker can email staff from that trusted name, exploiting workplace bonds to dodge suspicion [1]. A single stolen password feeds a wider black market [2].

Security frameworks separate online scams from other hacking tactics by tracking how an intruder first touches a network. The MITRE ATT&CK matrix lists drive-by compromises and phishing as two different entry tactics. In a drive-by hit, attackers plant code on websites that a target visits, using watering-hole schemes on hubs read by specific workers. Phishing, by contrast, needs a person to open a link or file, but crooks often point lures at watering-hole pages to drop malware. By pairing email lures with web drive-bys, hackers work around network boundaries and gain strong footholds inside corporate networks, linking two distinct entry paths into a unified intrusion strategy that catches users off guard [1].

Documentation portal for Microsoft Defender security tools and enterprise protections.
Microsoft technical documentation detailing endpoint defense strategies against deceptive credential harvesting campaigns. (Credit: Microsoft Learn)

Targeted Spear Phishing and Whaling

Bulk spam hits millions of random inboxes with identical text, but spear phishing customizes each note for one person or company. Crooks gather job titles, project names, and supplier rosters to make every line look authentic. Because these notes name real coworkers and active projects, workers drop their guards. When hackers aim these custom traps at chief executives, board leaders, or senior directors, the industry calls the scam whaling. Automation lets bad actors build these custom emails at low cost and high volume, combining personal research with machine tools to produce tailored messages that look indistinguishable from real internal memos. That level of customization drives click rates far above those of generic spam [1].

Controlled tests show how real people handle focused deception over time. In a field trial that sent fake spear-phishing notes to 100 younger adults and 58 older adults over 21 days, 43% of participants clicked at least one link. Older women logged the highest click rate among all four groups. While younger participants got wiser and cut their click rates over three weeks, older adults stayed just as likely to click throughout the entire test [1].

Targeted lures often seek enterprise cloud accounts [2]. Hackers frequently post fake calendar invites or deceptive online events designed to capture corporate login credentials to take over Microsoft and Google accounts. Taking one valid staff login gives an intruder an inside path to trick partners and vendors [1].

Voice Phishing Operations

Voice phishing, or vishing, uses phone lines and Voice over Internet Protocol (VoIP) to scam targets live. Crooks easily spoof caller IDs to display local codes or government numbers. Seeing a familiar agency name makes people answer and trust the voice on the line [7].

Criminologists have tracked how phone scams run like organized businesses. In Security Journal, Kwan Choi, Ju-lak Lee, and Yong-tae Chun ran a crime script study on 182 police files, 229 portal posts on South Korean search engines Daum, Nate, and Naver, and interviews with 20 detectives. Korean National Police Agency data showed over 4,000 cases of voice phishing each year, costing victims over US$1,000 per case. Their analysis mapped three stages: drafting scripts and hiring callers, making thousands of internet phone calls, and wiring out stolen cash. This systematic breakdown proved that phone fraud operates through organized gangs with clear job roles, separating callers from money-laundering couriers to insulate ringleaders from arrest [6].

Graphic published alongside reporting on digital deception and telecommunication scams.
Editorial graphic illustrating telephone network security concerns and telecommunication fraud. (Credit: The Register)

Callers prey on older people and immigrants with harsh threats. Pretending to be police or Internal Revenue Service agents, scammers threaten arrest or deportation unless the victim sends cash right away. Fraudsters also use robo-callers and recorded menus, pushing targets to dial fake help desks. Some rings now use cloned voices via audio deepfakes, prompting FBI alerts about VoIP fraud rings operating across borders [7, 10].

Fake links form the main path that pulls users into credential traps. Scammers set up deceptive subdomains, placing a real brand name ahead of a bogus domain (such as example.com.scam.com) to fool readers in a hurry [1]. Attackers also display a real web address in the email text while pointing the hidden link to an imposter site [2]. Some crooks even abuse open URL redirectors on trusted web pages to bounce visitors straight to a trap [1].

Internationalized domain names (IDNs) open another path through visual trickery. In homograph attacks, scammers buy domains with letters from other alphabets that look identical to normal letters in standard web browsers [1]. Microsoft security teams point out that scam sites often copy real portals by using flat screenshot images with floating login boxes. These fake sites show small flaws if you look closely: old logos, bad type, or strange boxes asking for personal info. Because casual readers do not inspect letter encodings while clicking links on laptops or phones, visual homographs fool everyday users into typing passwords into hostile web servers, handing full account access to unknown hackers without noticing any difference on the screen [1, 2].

Phones face growing risks from SMS scams, called smishing. Small phone screens cut off long web links, making it tough to check where a link really goes. The Federal Trade Commission noted that consumers lost US$470 million in losses to text scams in 2024, mostly through fake U.S. Postal Service package notices [1].

What is Phishing and How Do Attackers Steal Credentials? — arXiv.org image
(Credit: arXiv.org)

Adversary-in-the-Middle Session Hijacking

Newer tactics use adversary-in-the-middle (AiTM) setups to beat two-factor codes. An imposter server sits right between the user and the real web service. When the user types a login and a one-time code, the proxy passes them to the real site live, letting the attacker steal the open session [1].

Toolkits like Evilginx automate this theft by grabbing live session cookies. Cloud services use session cookies as bearer secrets to keep users logged in, so stealing one lets a hacker stay inside without knowing the password. Short-lived cookies help limit how long an attacker can roam [1]. Amit Klein, Chief Technology Officer at security firm Trusteer, found a related path called in-session phishing, showing how browser JavaScript bugs let rogue pop-ups trigger inside online banking tabs [11].

Physical items also lead into digital traps through QR code scams, known as quishing. Because square QR codes hide their real web links from human eyes, people cannot easily tell safe codes from fake stickers pasted on parking pay meters. In lab trials with 1,876 people, almost no participants spotted fake QR codes [1]. In IEEE Networking Letters, Muhammad Wahid Akram, Keshav Sood, Muneeb Ul Hassan, and Basant Subba showed how combining QR codes with Browser-in-the-Browser (BiTB) traps created dangerous lures. By testing bad prompts on the Google Gemini LLM API and comparing Gemini with GPT and plain sites, they proved that smart chatbots can speed up credential-theft attacks against end users [5].

Cover publication graphic from SpringerLink for academic research into voice phishing.
Springer publication archiving empirical research into voice phishing operations and crime script analysis. (Credit: SpringerLink)

Generative Artificial Intelligence in Social Engineering

AI tools have changed how scammers plan and write their attacks. Marc Schmitt and Ivan Flechais shared their Generative AI Social Engineering Framework in a 2023 preprint on arXiv, noting their ideas had not yet seen peer review before publication in Artificial Intelligence Review [3, 4]. Their model highlights three pillars where smart models boost attacks: generating lifelike content, tuning lures to specific targets, and building automated attack setups. These pillars explain why modern language models make online scams far harder to detect [3].

Staff training once told people to watch for typos, bad grammar, and odd greetings [2]. Today, large language models wipe out these old clues by writing crisp, fluent text fitted to any work context [3]. In a 2026 test of 7,700 people, AI-personalized emails drew nearly three times the click rate of generic scam emails. The cost was tiny: about US$0.03 per message. That cheap efficiency lets attackers run customized schemes at scale without hiring writers or spending days researching targets [1]. Because machine models generate natural workplace memos and clean requests, employees can no longer rely on grammar checks to spot fraudulent mail, forcing companies to adopt strict cryptographic sign-in methods that do not depend on human judgment alone [2, 3].

Bots also help set up attack pages and web code. Scripts build fake web pages and synthetic images fast [3]. As firms face automated artificial intelligence attack workflows, keeping systems safe means using automated defenses that react at machine speed [2].

Historic Breaches Across Three Decades

Phishing traces back to early online consumer services in the mid-1990s. In 1994, the AOHell program let scammers pose as America Online workers via instant messages to ask users for passwords. By 1995, AOL account theft was common, and the word phishing showed up in print by March 1997, drawing its name from phone hacking subcultures called phreaking. The first recorded phishing strike on cash systems hit E-gold in June 2001, followed by a post-9/11 ID check scam. Retail banking customers faced hits for the first time in September 2003, setting off an era of organized online bank scams [1].

Money losses climbed fast as gangs commercialized scams. Between May 2004 and May 2005, roughly 1.2 million U.S. users lost US$929 million [1]. In the United Kingdom, web banking fraud losses nearly doubled in 2005 over 2004 [9]. In 2006, half of global phishing ran through the Russian Business Network in St. Petersburg, while scammers hit IRS taxpayers and Myspace users. In 2007, 3.6 million adults lost US$3.2 billion [1].

Targeted hacks hit major groups in the 2010s. In 2013, thieves phished a heating contractor to breach Target, taking 110 million records [1]. In 2014, hacker Ryan Collins breached over 120 Gmail and Apple iCloud accounts, stealing photos of Jennifer Lawrence. Collins pled guilty under the Computer Fraud and Abuse Act. United States Attorney Eileen M. Decker stated: “People store important private information in their online accounts and in their digital devices. Unauthorized access to such private information is a criminal offense.” [8] Russian group Threat Group-4127 (Fancy Bear) later phished the Democratic National Committee in 2016 [1].

Technical Controls for Enterprise Defense

Defending systems needs strong machine controls, not just staff training. Companies deploy tools like Microsoft Edge and Windows Defender Application Guard, using Hyper-V virtualization technology to run untrusted sites in isolated containers. These containers seal off bad web code from reaching files or internal networks [2].

For cloud work, teams run Microsoft Defender for Office 365 across apps like Microsoft Teams, Word, Excel, SharePoint Online, and OneDrive for Business. The tool checks risky files and scans web links live to give zero-day defense [2]. Because proxy setups steal passwords and texted codes with ease, IT leaders push for phishing-resistant keys. Tying logins to hardware tokens stops stolen codes from working on other machines [1].

When a worker suspects a scam, fast action cuts down harm. Users on work PCs should alert IT staff right away, reset all account passwords, and call their banks to freeze credit cards. People can report scam emails inside Outlook by opening the ribbon and choosing Report Phishing, while bad websites can be flagged inside Microsoft Edge. Samples sent to the Anti-Phishing Working Group (phishing-report@us-cert.gov) help internet providers, banks, and police shut down scam web hosts. Through quick reporting and shared threat data, security teams block rogue infrastructure before attackers gain permanent administrative footholds across core enterprise infrastructure, ensuring that compromised accounts are isolated and neutralized before widespread damage can occur [2].

Sources
  1. WEBSITE Wikipedia contributors. (n.d.). Phishing. Wikipedia, The Free Encyclopedia. [Article Link]
  2. WEBSITE Microsoft. (2026). How to protect against phishing attacks. Microsoft Defender for Endpoint. [Article Link]
  3. PREPRINT Schmitt, M., & Flechais, I. (2023). Digital deception: Generative artificial intelligence in social engineering and phishing. arXiv. [Article Link]
  4. ACADEMIC JOURNAL Schmitt, M., & Fléchais, I. (2024). Digital deception: generative artificial intelligence in social engineering and phishing. Artificial Intelligence Review, 57(12). [Article Link]
  5. ACADEMIC JOURNAL Akram, M. W., Sood, K., Hassan, M. U., & Subba, B. (2025). Exemplifying Emerging Phishing: QR-Based Browser-in-the-Browser (BiTB) Attack. IEEE Networking Letters, 7(4). [Article Link]
  6. ACADEMIC JOURNAL Choi, K., Lee, J., & Chun, Y. (2017). Voice phishing fraud and its modus operandi. Security Journal, 30(2), 454–466. [Article Link]
  7. WEBSITE Wikipedia contributors. (n.d.). Voice phishing. Wikipedia, The Free Encyclopedia. [Article Link]
  8. ONLINE NEWS Kamps, H. J. (2016). Prosecutors find that ‘Fappening’ celebrity nudes leak was not Apple’s fault. TechCrunch. [Article Link]
  9. ONLINE NEWS Richardson, T. (2005). Brits fall prey to phishing. The Register. [Article Link]
  10. ONLINE NEWS Leyden, J. (2008). FBI rings warnings over VoIP phishing cons. The Register. [Article Link]
  11. WEBSITE Wikipedia contributors. (n.d.). In-session phishing. Wikipedia, The Free Encyclopedia. [Article Link]
Leave a Comment

Related Posts
Total
0
Share